By: Peter Williams, Practice Leader - IT Infrastructure Mgmt., Bloor Research Posted: 24th July 2008 Copyright Bloor Research © 2008

I gained a more positive view of post-process de-dupe—or rather what I would call ‘partial post-process'—from meeting with virtual tape library (VTL) appliance provider SEPATON last week. Its new DeltaStor de-dupe approach is unique and so deserves a separate review.

De-duplication performed during an initial backup—‘in-line' so called—is typically achieved transparently to any management process by an appliance and (with compression included) can achieve a 20x or more space saving over a standard backup. Applied across the board to every file and system, it typically treats them all just as blocks of data without taking account of file type or content. ‘Post process', which applies de-dupe to a backup only after it is created, initially requires extra space and typically incurs some management overhead; this is not so smart in my book.

SEPATON's DeltaStor is technically ‘post process' but different. Its software examines the backup copy of each individual file and database (‘object') in turn but, uniquely, uses its ContentAware stored intelligence to recognise all the leading vendors' backup and archive output as these embed their own markers. In SEPATON's de-dupe process these markers are extracted before the data is processed.

There then follows a byte-level examination of the whole data stream; from this the de-dupe process (which does not use hashing) creates variable-length output representing anything from 128 bytes to the whole object. "Nobody else does that," said Miklos Sandorfi, SEPATON's CTO, who pointed to a verified 48x space-saving typically being achieved in its VTL output. It still needs additional space but, as Sandorfi explained, far less than you might expect...

Since each backed up file or database is handled as a separate entity, DeltaStor can be set to start work on de-duping the first file as soon as that backup is complete and concurrently with the next file backup, and so on (so effectively only ‘partial post-process'). This also means the minimum amount of output space that has to be pre-allocated is the size of the largest file to be backed up plus the total de-dupe output space (which all de-dupe products need); then remember that DeltaStor's de-dupe space will come out less than half that used by the best in-line de-dupe products. Some files should not be de-duped (for instance already encrypted ones); with DeltaStor's approach the decision whether to de-dupe can be set at the most granular single-file level to further assist space-saving.

So when calculating the overall space saving versus the best in-line solutions, consider: a) the total amount of data to be backed up (typically more for larger enterprises), b) the degree to which further replication is to be applied to the de-duped backups (since with SEPATON these instances will be smaller, which also helps performance especially if some travel over a WAN), c) the effect of some files not being de-duped, and d) how long the data is to be stored accessibly from disk (since the longer it is retained in this near-line de-duped state the bigger the space-saving).

Logically, SEPATON's approach is most attractive to larger enterprises with larger and more complex backup and archiving needs who do not mind a minimal amount of extra management. In exchange SEPATON offers some enterprise-level additions.

For instance, there is a rigorous byte comparison check on data integrity. Sandorfi says that SATA disks have a habit of changing the data without showing an error. (Very nasty if true!). Also, SEPATON's ‘forward differencing' approach reverses the way most de-dupes work. Whereas they use the first instance of data as a reference copy and replace subsequent instances by a pointer, DeltaStor stores the most recent data copy in full form—replacing old and redundant data with pointers. This circumvents two problems: a) a gradual tail-off in backup performance and b) a delay in restoring the most up-to-date data.

In-line solutions that cannot maintain wire-speed will impede initial back-up throughput performance. Although not a like-for-like, SEPATON's VTL appliance does boast up to 34.5TB/hour as well as scalability to 1.6 petabytes of data.

Finally, through its software being aware of the content, SEPATON is working to develop other functionality, for instance to facilitate secure, audited content searches for legal discovery. (But that is for another day.)

Right now the decision for in-line or SEPATON-style partial post-process depends on organisation size and needs. But I still await a convincing argument for standard post process de-dupe.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Peter Williams (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Peter Williams, Practice Leader - IT Infrastructure Mgmt., Bloor Research Posted: 21st July 2008 Copyright Bloor Research © 2008

My focus is on storage and infrastructure management so it is unusual for me to review new network-based equipment. What interests me about the ONStor Cougar announcement last week is how it might impact enterprise storage management.

ONStor's Cougar is a NAS gateway storage solution; primarily it doubles (and more) the capacity and throughput of its own Bobcat predecessor. As with the Bobcat, Cougar units can be sold purely to form a clustered NAS gateway fronting other companies' storage arrays, or with its own storage arrays attached. ONStor's gateways already front storage arrays from most vendors, the favourites being Nexsan, Engenio and Hitachi.

A gateway's performance is crucial. It can be a bottleneck to overall storage throughput but Cougar can cope with a huge 3MB/s (double that of Bobcat). Its heterogeneity to handle any array or protocol means, from a management perspective, that the overall network is largely unaffected even if the storage arrays behind the gateway are constantly being swapped out for more capacity.

Looking at raw storage capacity, both Bobcat and Cougar can start with a mere 2GB then expand incrementally—Bobcat to 1PB but Cougar on up to 4PB. In market terms, these capacity and throughput performance figures mean Cougar has broken loose from ONStor's mid-range roots into enterprise-level capacity scalability—to attack the likes of long-time suppliers such as Network Appliance (NetApp) and EMC.

The way the raw performance figures are achieved is interesting, at least to the techies. The clustered NAS filer uses a 64-bit pipelined architecture with 2xQuad MIPS processors plus the operating systems (achieving a maximum of 18 cores per system) with 8Gb/s Ethernet or 8x4Gb/s fibre channel ports and a TCP offload built in. Memory is currently 12 or 16GB (expandable to 32GB in future). As you would expect, there is full hot-swap redundancy built in—and all this in a compact 2U chassis.

What this means in practice is 840 MB/s sequential read gateway performance (basically the 8Gb/E line rate) and a still outstanding 700 MB/s sequential write (array-limited)—benchmarked to 101,000 SpecSFS operations per filer. The company claims this means it achieves 1.5–2 times the price-performance of all competitors and with linear performance scaling.

Other features include PCI Express (native QLogic or Connectix) and the ability to include data de-duplication, encryption and compression (all of which are becoming standard requirements if used in a back-up and archiving environment at least).

But what may also help turn the heads of enterprises is the surprisingly low energy consumption—300W for the 2U unit—and space-saving. ONStor says Cougar achieves 4x the space and 3x power and cooling savings for the same data capacity; the latter is partly achieved through using MIPS processors which have evolved to achieve a lean, efficient pipeline that uses less than half the power of an x86.

All of this needs to be set in the context of rocketing data storage needs—and what that means for IT management. A first thumbs up is that I see little reason for a Cougar NAS gateway to upset existing management procedures; yet it could well provide an immediate fix to storage throughput problems affecting SLAs. A by-product may be that faster data streaming applications can be handled for the first time in some businesses.

There is also a good green (as well as cost) message; wow, the world's first green Cougar! The NAS gateway with storage could immediately address limitations in a data centre with space, power usage, heat dissipation and so on—and reduce ongoing running costs. This can be taken further by the small incremental storage expansion capability; this can keep unused capacity to a minimum (without even deploying virtualisation or thin provisioning), again minimising power and space problems.

ONStor means business as it sees enterprises engaged in rapid globalisation as being prime targets. The company told me it was seeing customers adopting the largest clusters they could get, partly to get five 9s availability; it has one customer with 28 Bobcats.

Taken together these factors do provide a positive management message—and I will be watching with interest to see how the storage vendor "big boys" react.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Peter Williams (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Peter Williams, Practice Leader - IT Infrastructure Mgmt., Bloor Research Posted: 14th July 2008 Copyright Bloor Research © 2008

Storage de-duplication has the potential to be used in lots of situations—and de-dupe specialist Data Domain is having to work hard to prioritise provision of new features from the opportunities it is seeing.

The starting point is using its NAS-style de-duplication storage appliances which can be installed with minimum disruption to an organisation's existing way of working. This means that, for instance, it carries out an in-line de-dupe transparently within an unchanged backup procedure. The company says this will typically achieve an immediate 20x backup disk saving and requires no management.

So my question is: "Why wouldn't you?" Yes, you have to pay for the de-dupe appliance but the massive disk capacity savings achieved means avoiding future disk drive purchases. In turn this can, for instance, greatly defer the day when your data centre runs out of capacity (space, energy) so it also fits well with a green IT policy.

Data Domain also uses this de-dupe process for a virtual tape library (VTL). The huge disk capacity saving means data can be economically retained on disk—nearline storage—for, perhaps, months before there is a need for it to go into deep tape (or optical) archive. In the meantime it is much more rapidly recoverable and accessible. With the data taking, say, 1/20th the capacity on low cost SATA disk compared with ‘un-deduped’ tape, the economics of disk versus tape is radically altered in disk's favour.

In both cases the data is accessible reasonably fast, so it provides a nearline tier which can be accessed directly for many applications; for instance Data Domain has partnerships with a couple of content search engine providers. Storage content searches are useful as input to discovery as evidence for a compliance court case.

A new Data Domain feature is Retention Lock; this can set a lock on individual files as they are archived so that they cannot be changed in any way for a pre-set period. Since this is open for the IT manager to set or change it is not suited to rigorous SEC-level compliance, but helps ensure good governance since it will firmly block user access. The company also uses a partner to provide encryption. Together these steps show Data Domain making at least tentative moves into accommodating governance, risk and compliance (GRC) needs. A data destruction verifiable delete facility is also planned this year.

In fact de-dupe is equally at home with archiving as with backup, although the nature of archiving means the space saving of, typically 75–80% or 4x, is much lower than for backup; but it's still impressive. Moreover, the process is also helping remove the demarcation between backup and archive systems which, at least longer term, should help simplify the management process.

Further ways this is supported is that sending either a backup or archive copy to a remote location, even travelling over a WAN, is practical. Now add a frequent snapshot capability which sends hardly any data as it only needs to store data tags, and you nearly have continuous data protection (CDP) and a very low-cost disaster recovery (DR) solution. You also obviate any need to physically transport newly-created tapes to a remote secure location—by sending the information over the wire.

All these are possible only because the specially-designed appliance, which draws heavily on CPU performance, achieves the necessary throughput to carry out block- and byte-level de-dupe in-line as the data is received. Any vendor providing only a software solution cannot achieve this throughput—and building an optimised appliance is not an overnight job. The alternative, so-called ‘post-processing’ de-dupe that only works on the already backed-up storage, has very little value in my book, as it needs to allocate more disk space and incurs extra management.

So, notwithstanding the economic downturn and with storage volumes set to continue soaring, Data Domain looks to be sitting pretty right now.

What of the future? Clearly, since applications can already access de-duped nearline storage in real time, there are few technical reasons stopping de-dupe being applied to tier one (even tier zero) storage and saving yet more space—except in considering when to accomplish the de-dupe. (No immediate plans for this I'm told.) What I do know is that Data Domain's own users are thinking outside the (storage) box to pass on their ideas—so some highly original future developments are entirely possible.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Peter Williams (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Peter Williams, Practice Leader - IT Infrastructure Mgmt., Bloor Research Posted: 29th May 2008 Copyright Bloor Research © 2008

It was good to see the Storage Network Industry Association (SNIA) Academy finally make footfall in the UK, with its all-day event last Tuesday in London.

The Academy has been staged in locations around Europe since 2005, and the emphasis is on education, especially about the latest storage technology, and the trends, challenges and issues. The material throughout the day was high quality and hit the major topical themes those involved with IT storage would want to hear more about. So well done SNIA.

I can only skim the surface here and it is only my take anyway. So my apologies to those I fail to mention (or even hear, as there were break-out sessions running in parallel and I could not attend everything).

Jon Tate, SNIA Europe UK committee chair, opened the proceedings by saying the Association had been in transition since 2004–5, emphasising not just storage but also information—an obvious move since all stored data is information and, increasingly, decisions on storage need to be made based on what the actual information consists of.

Then followed a very sobering presentation from Ann La France, worldwide legal counsel for Squire, Saunders and Dempsey, who knows a thing or three about the thorny subjects of compliance, and data protection versus freedom of information. One of her themes was data retention (backed by the EU data retention directive of 2006). She pointed out that the best solution to protect against data breaches was to delete the data after the minimum time period. The average cost of a breach, with the loss of unencrypted data was estimated at £45–£70 per record—just in lost business and administration—while the loss of consumer confidence was unquantifiable, she said.

The two opposing forces pulling against one another were privacy encouraging an early purge and governments wanting to access personal data because of national security concerns—with the UK government currently in breach of EU directives in this! In the middle sits regulatory compliance requiring certain information but also demanding security including deletion. Frankly, nobody much is deleting anything at present, and La France thought many were ignoring the problem hoping it would go away. Meanwhile the storage mountain grows.

It might have been useful to put La France in a panel debate alongside Nick Baker of Sun Microsystems, whose theme was ‘best practices for long-term retention of digital information'. He qualified this title with the word ‘preservation'—pointing out a major problem of retrieving long-held data. SNIA had carried out a 100 year archive survey. Frighteningly, 68% of the companies contacted have data they say needs retaining over 100 years rising to 83% over 50 years. 53% even said they had data needed in perpetuity.

In some cases this longevity stemmed from requests by government. So shouldn't governments defray the costs? (Oh, that means the tax-payer pays; perhaps I should retract that.) Preservation, said Baker, was a bigger problem for semi-structured or structured data; for instance, Oracle objects and tables relate to each other so metadata is needed to describe the information stored to make it genuinely discoverable.

Apart from a regular technical refresh involving migration to latest software versions there was the matter of physical and logical migration as formats became out of date. Baker emphasised that logical and physical should not be mixed—and, he said, only some 30% were doing this correctly on disk while nobody was for tape or optical. In other words this was: "record to tape and lose."

SNIA's answer was a ‘holistic approach', not stove-piped with silos of uncorrected information, which required an understanding of what an object was in every case. The metadata format had to be correct and an audit trail maintained from the original object with an archive object versionary needed.

He also put in a plug for SNIA's XAM emerging standard for metadata (which I have previously covered and believe has longer-term potential).

Despite this, it all sounds expensive and time-consuming to me. Worse, said Baker, it was at the bottom of the IT hierarchy so lacked adequate funding, therefore should be pushed back to business as a serious risk.

Among some of the other main items was John Rollason (SNIA UK committee and NetApp) covering every aspect of storage virtualisation and how to use it effectively and Bill Bolton (SNIA UK and Brocade) giving us just about all we should ever need to know about Fibre Channel, its history and clear road ahead. Mark Galpin of Quantum's overview of de-duplication technologies highlighted major differences in de-dupe approaches, while Steve Collins of Pillar Data Systems covered various current trends in data protection and restoration technologies, not least CDP.

The final presentation of the day, by Sol Squire (SNIA Europe Nordic Committee and Data Islandia) on building a green data centre, was full of practical tips for data centre managers, overwhelmed by their challenges. Not the least of these was spending a little money on data centre sensors so as to plot the power flow in the data centre. "60% of cooling is wasted; measure what you have," he said.

Illustrating the point he told of data centre managers identifying the flows then strategically placing a shower curtain to save 40% of the cooling bill at a stroke! On a similar theme of heat output versus cooling, he said (perhaps to the consternation of some company security managers), "You can open a window in the data centre." (The ultimate alternative of building a new data centre when resources run out has an average cost £20m.)

Squire also advocated investigating renewable energy. (Iceland, where he is based, runs on 100% renewable energy, but only has 300,000 population.) He also recommended having small realisable goals as little things had greater effect down the line. Then, he said, "hopefully our grandchildren will still look up and see a blue sky."

Finally, couple of points from two vendor-specific break-out sessions I attended, are worthy of a mention.

Trevor Kelly, EMEA systems engineering manager for 3PAR, was discussing thin provisioning. In the course of this he cited a recent Glasshouse Technolies' survey of 350 host systems in 12 large companies - which showed storage utilisation still below 30%. Frankly, with the virtualisation and other technologies now available and the green impact moving up the agenda, this is now an unacceptable waste of resources.

Meanwhile, Rick Terry of IBM provided interesting - nay, alarming - slides about how disk areal density improvements which had for decades kept pace with Moore's Law for computer chips - were now tailing off. So, he predicted a disk price crunch as it was going to be more difficult to get larger capacities - and, with the huge data capacities now needed, small error rates extrapolated to more frequent failures. So, he said, "A 1PB (petabyte) drive fails every 10 days."

Maybe there's an overall message on the day: Try and tackle the storage mountain itself and do some serious data deletion. That way, all the other issues and concerns will reduce in size and cost.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Peter Williams (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Peter Williams, Practice Leader - IT Infrastructure Mgmt., Bloor Research Posted: 12th May 2008 Copyright Bloor Research © 2008

De-duplication (de-dupe) specialist DataDomain today [12th May] launched its fastest—and the industry's fastest—de-dupe appliance, the DD690.

The stand-alone unit, which has an Intel quad core CPU, can achieve up to 1.4TB/hour throughput, or up to 170MB/sec in a single-stream for a large database. Its addressable physical capacity is 35.3TB. However, up to 16 DD690 arrays can be installed in the DDX cabinet, to provide a maximum capacity of 28PB and an aggregate throughput of 22.4TB/hour.

What DataDomain describes as "the fastest single-stream de-dupe engine in the world" uses its own operating system with an architecture to maximise processing and indexing within the CPU while minimising disk access. DataDomain's VP of product management Brian Biles told me that this approach would be difficult to match. "We expect a 50% de-dupe speed-up every time Intel doubles the number of CPUs."

The big positive in de-dupe is that it drastically reduces nearline and/or offline storage. (Primary storage is never de-duped.) In simple terms, all the data is examined as it is backed up and duplicate files and/or blocks (or even smaller segments) are replaced by a pointer tag to where the single instance of the data is held.

This picture becomes even more attractive—at least using the DataDomain architecture—if data is coming from a number of remote locations. A further announcement is the extension of its processing capability to allow up to 60 separate data streams to be de-duped before transmission over a WAN to, for instance a single large receiving DDX. Clearly, the less physical data that needs to travel over the wire, the faster this is completed. Moreover, bringing separate data silos together in this way means a greater space saving—since, otherwise, identical data will remain replicated in the different silos—and is easier to manage centrally.

Competitive approaches are, in general, not as efficient in space-saving; but DataDomain expects its de-dupe to achieve a 10x reduction and, combined with data compression, will achieve a 20x output space reduction—with a 95–99% reduction in cross-site transmission bandwidth. Then comes the question: "In that case why wouldn't everyone do this?"

There are several answers to this, and most of them revolve around where it is applied. DataDomain's phenomenal growth—doubling turnover in three quarters since it IPO'd last year and adding nearly 300 new customers in the past quarter alone—is a testament to the fact that most large organisations are now taking the plunge. (The company is now the de-dupe market leader in the US and a recent survey suggests it will soon overtake Symantec and EMC in Europe too.)

A second deterrent is that de-dupe performance can be a dog. That is why throughput figures are key, and why a stand-alone de-dupe appliance tends to be the best option; this is also part of why some vendors only provide de-duping on already backed-up data (as a background task that still slows down other processing and saves less space). In some systems performance can be improved but only at the expense of output disk space, to some extent defeating de-dupe's whole purpose.

Perhaps most of all it is considerations about recovery from backup—and especially disaster recovery (DR). The data is not a mirror of the live system and, if recovery is needed, it has to be ‘reconstituted' to its original form. So, if an immediate switchover to a remote DR site is needed, forget de-dupe—at least for the critical on-line data. But in all other cases, it comes down to how efficiently and reliably the recovery can be performed.

The output file will be the same in either case. Retrieving blocks using existing pointers will be faster than when they had to be created; but, of necessity, it will be slower than contiguous reading—unless segments are placed in optimum sequence and sometimes memory-held so no disk access is needed at all. DataDomain does try to arrange its stored segments optimally for retrieval performance.

Potentially, then, recovery can be as quick as would be the case without de-dupe—largely removing that objection. Likewise, if some de-duped data goes to off-line archive on tape, and has to be reconstituted first. However, this performance level will not be achieved by all solutions, so performance testing vis-à-vis likely SLA requirements should be considered in all major evaluations.

The reliability and integrity of the data, and protection against hardware faults, is covered by such things as on-the-fly error detection and correction and continuous disk ‘scrubbing' to remove errors before they become a problem. So, short of a DR regime where up-to-the-minute mirroring is crucial, I ask again: "why wouldn't you include de-dupe?" It is, anyway, very green (greatly reducing disk space and network bandwidth), so energy and heat saving—which provides an obvious ROI.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Peter Williams (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Rob Bamforth, Principal Analyst, Quocirca Posted: 22nd April 2008 Copyright Quocirca © 2008

The importance of assessing and measuring records management policy and the subsequent procedures used has been thrown into sharp relief by the disaster that last year befell the UK's Revenue and Customs organisation. The personal records of 25 million of the UK population were sent protected only by a password in an internal postal system, and lost.

It's bad enough that it happened at least once, but worse still to think that organisations get away with bad procedures for some time, before only a combination of circumstances cause them to be revealed. Even if the strategic policy is robust, when a system breaks down in the implementation of the policy, its effectiveness isn't being correctly measured.

Organisations need to know how their systems and processes are performing—well or badly—before any fallibilities are exposed with potentially far more catastrophic results. There is also the more basic and straightforward reason for checking—having invested in a records management system, is it providing good value?

Most organisations will have justified the expenditure on some combination of expected benefits—strategic and tactical. In both cases, they need to know if these benefits are being realised and must put some form of measurement and benchmarking systems in place.

There are two main drivers for this. The first is to demonstrate the short term benefits to stakeholders to show where the investment has been a success, and where there needs to be changes or enhancements based on lessons learned. It should not just be a set of check boxes as part of the project review and seen by only a few, but a broad internal presentation of current and expected progress. This should have been set out as a main part of the initial plan, and have straightforward measurement criteria set against each benefit, including:

• improved productivity
• competitive gains
• costs savings
• reductions in storage space
• better workflow

The second reason for measurement is for the longer term, more strategic or indirect benefits that need to be enhanced and protected. This includes those that might form part of an external vision or message, both for ongoing present promotion, but also for defence in the event of some future failure. These benefits are harder to quantify, but should also have been outlined from the beginning:

• image or brand value
• the need to meet statutory requirements
• enhanced knowledge management
• improved customer service
• business resilience or disaster recovery

How should an organisation start to quantify what they might need to measure in their records management projects? Well there are established standards to address this, such as ISO 15489, which covers plenty of ground in addition to providing a benchmark for best practice.

ISO 15489 is about the entire approach, methodology and processes for ensuring that an organisation's records are properly managed and made usable and accessible throughout their lifecycle. For the sake of external validation and verification, the standard also ensures that critical stages, such as final disposal, are carried out in an open and transparent manner and according to pre-determined criteria. This is particularly important where there are regulatory or data protection requirements from legislation.

The value of measurement is not only for the benefit of external stakeholders. Improvements made in information access and workflow can have a huge impact on individuals, both in terms of their effectiveness and their job satisfaction. Office workers can waste many hours searching for badly labelled, badly filed or simply ‘mislaid' records, which eats into their time and morale. So as part of the introduction of new procedures and systems, the personal benefits can be identified, and then captured over time to show even the most reluctant individuals and, perhaps, more importantly, their industrial relations representatives, that improvements are being generated for both organisation and individual.

Tackling measurement from an early stage can also help if a project is struggling to obtain sufficient management resources by identifying some early success stories. If some of the expected benefits have not been achieved there will be a need to look for reasons and seek to overcome them so that later phases can be directed to address the shortfall.

However, it is best not to over-rely on measurement, and more useful to measure what is important, rather than making important what it is often too easy to measure. Always feed back good news items with reasons as to why they have happened and how they will be capitalised upon, but don't shy away from making clear where there are problems, and what corrective action will need to be undertaken.

Finally, if internal measurement or external exposure in the media highlights major flaws, take prompt action to tighten up procedures and communicate them so that all staff understand why processes have become more restrictive. These can always be relaxed later, but everyone needs to know that the organisation takes its records management seriously and effectively measures how well it is being performed. No one wants to discover inadequacies after a public failure—just ask the UK's Revenue and Customs, or any of the other large organisations that have recently inadvertantly hit the media.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Rob Bamforth (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Clay Ryder, President, Sageza Group, Inc. Posted: 17th April 2008 Copyright Sageza Group, Inc. © 2008

IBM has introduced the IBM Power System, the first of a new generation of servers unifying the former System i and System p product lines, which features simplified pricing and increased application choice as well as reduced energy and administration expense. As part of the new launch, IBM's integrated operating system formerly known as i5/OS will now be known as ‘i'.

Key features of the new Power System family include:

• POWER6 processors
• POWER6 EnergyScale Technology that offers advanced energy control features
• PowerVM virtualization technology support for up to eighty virtual partitions
• Simultaneous support for AIX, Linux and i applications

There are three initial offerings from the new product family, which are targeted at SMBs:

• i Edition Express for BladeCenter S targets existing AS/400, iSeries, and System i 515, 520 and 525 customers that are looking to refresh or extend their existing investments as well as simplifying the integration of i applications with x86 servers. Customers can optionally migrate existing x86 servers onto x86 blades in the same BladeCenter chassis for increased integration and simplicity.
• The IBM Power 520 Express is designed for businesses running distributed applications, databases, and core business solutions.
• The IBM Power 550 Express is directed at organizations that seek very high performance and capacity in a mid-sized database server while also requiring continuous application availability.

The new Power servers are available in an i edition, AIX edition, or Linux edition. Each edition includes the server, packaged components, and the operating system. These editions offer customers a preconfigured offering that enables quick deployment. In addition, customers can also order a-la-carte to mix and match i, AIX, and Linux on a single Power server based upon their specific needs.

With this announcement, IBM Business Partners holding System i and System p certifications will be able to sell a combination of solutions and operating systems on the new unified Power platform. ISVs will have a single platform on which they can develop i, AIX, and Linux solutions, potentially broadening the audience for their applications, especially in multi-OS datacenters.

In a separate announcement, IBM Global Financing unveiled several IBM Power Systems financing offerings. Among the offerings are a "one platform, one monthly lease price" total solution financing for Power Systems that includes the hardware platform, OS, peripherals, and maintenance bundled in a single price. In certain geographies, there are special lower-rate financing and deferred payment programs available. In addition, mid-lease upgradeability for little or no change in monthly payments may be available for customers seeking to acquire the new technology.

Net/Net

This is one announcement that we have been waiting to see for a long time, and we are elated that it has finally happened. While the continued gains in IBM's POWER6 processor are noteworthy, and its unique position in the marketplace as a highly flexible multi-OS 64-bit platform should not be underestimated, what captures our attention is the unification of the former System i and System p product families. To our way of thinking, this unification bodes very well for the UNIX community and hopefully will put to rest the ongoing angst of whether or not IBM truly supports and believes in the beloved System i. To be clear, in this announcement System i aficionados have much to be thankful and a lot to look forward to.

While many of the advances of the Power processor family have made it to both the System p and System ii product lines, the absence of an i5/OS-supported blade has been a rather obvious omission. Happily, with this announcement, organizations that are serious about consolidating UNIX, Linux, and i5/OS applications into a simplified solution can now do this very task in a state-of-the-art blade environment. Since the BladeCenter S solution for i has starting prices that are at about half the price of comparable non-blade configurations, this new value proposition may prove attractive to customers that are still operating older System i, iSeries, or AS/400 servers but have had a difficult time justifying equipment refresh that may have simply meant a newer, faster server running the same applications.

The recognition that the value proposition of i is the software stack is a realization that we believe is long overdue. The longevity of the applications deployed on these environments is testimony to the load-it-and-forget-it simplicity that is so appealing to SMBs. With the new Power Systems, existing System i customers can increase performance and energy efficiency by migrating their mission-critical applications to new systems (standalone servers or blades) and continue to use the same i applications and operating system. For such organizations, the ability to embrace the future while consolidating the present may be the missing piece of the ROI puzzle that justifies new IT investment. Ultimately, for i, it's not about the hardware, it's the applications, and the administrative and user experience; all of which we believe are safe and potentially further enabled through the unified approach.

When one considers the processing agility of the POWER6, many workloads could be consolidated whether they are Web-oriented infrastructures based upon Linux and open source, i5/OS, or AIX. Through the PowerVM Lx86 (former System p AVE) developers could leverage existing x86 Linux apps or even continue to write Linux apps on their x86 workstations, and then take advantage of the scalability and efficiencies inherent in the POWER architecture when it comes time to deploy the app. This capability can also benefit ISVs as they instantly have a larger addressable market, especially in the higher end of performance needs, and can simply deliver their existing product and expertise for another platform.

For customers seeking native ports, this offers a stopgap by which the customer can deploy existing applications on Power and then swap out for native ports as they become available. Alternatively, given the native x86 support in the BladeCenter S, Linux applications could be moved from x86 to Power and back to x86-based blades as business requirements dictate. The benefit to VARs and SIs is the same, the option of selecting from thousands of additional applications to weave into a customer solution. With the new unified platform, VARs and SIs will likely have an easier time focusing on the customer solution and not explaining the differences between the historically different software stacks on the Power platform.

Overall, we are pleased with the scope and nature of this announcement. The value of Power as a cutting edge UNIX and Linux application platform remains evident. The heritage of the System i is assured and its existing customer base's loyalty has been rewarded. The Power System i Edition will continue to offer the integrated computing solution on which so many SMBs rely while at the same time, customers will be able to benefit from lower prices for server components and now have access to leading-edge blade environments on which to run all of their Power-based applications. The value of these applications is being redefined as more relevant than ever to a marketplace that is crying out for the operational and management ease of use typified by i environment.

For Power platform users, it's been like an extended family holiday gathering. Some years some can make it, and others cannot. With the new Power Systems, we are happy to see that all of the siblings in IBM's Power platform i, Linux, and UNIX user communities will finally be able to come together under a unified platform while each maintains its own highly valued business propositions. This is a family reunion that would be worth attending.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Clay Ryder (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Peter Williams, Practice Leader - IT Infrastructure Mgmt., Bloor Research Posted: 15th April 2008 Copyright Bloor Research © 2008

Some of us interested in full information lifecycle management (ILM) have long pointed to a need for an industry standard format for metadata that describes the data content of files to a more granular degree. Currently, software using metadata in this way has first to create it to its own proprietary format—which is typically unusable by any other vendors' software wanting to access the same data.

This being primarily a storage problem, the obvious body to drive development of a standard is the Storage Network Industry Association (SNIA). SNIA's response has been the development of eXtensible Access Method (XAM) specification, version 1.0 of which was released last week. The SNIA has yet to gain member approval but, assuming this is achieved by mid-year as intended, it will then submit the specification to ANSI and ISO for accreditation. Mid-2008 should also see release of the XAM SDK available under licence for industry developers.

Hold on a moment. I understand XAM is only addressing fixed format content at this time. This, though, is not the biggest problem. It may be a little inefficient but a user can, if needs be, hard-code access to a particular file type if the format of the individual fields is known; the code can then do field checks and be as granular as needed to decide to which storage tier to assign the data or move it—even without creating metadata.

Don't get me wrong, I can understand that creating a standard metadata mapping to the important fields of all fixed format files using standard syntax means standard routines can be used instead of reinventing the wheel for each file. If the process is not made too complicated and long-winded to set up and inefficient and slow in use there is good reason to see XAM adopted over time.

Yet the bigger problem is handling free-format content. This is necessary not least because the increasing regulatory burden includes maintaining documents (including e-mails and, soon, voice-mails) which contain free-format text. Software generally ducks the problem of looking at the content of these files as received, creating metadata for them, and assigning them to appropriate storage tiers and—most importantly—properly managing it so that the vast majority can be moved to low-cost off-line storage in a matter of weeks. (A few vendors, notably Njini, have tackled this.)

Instead, organisations keep the data for years "just in case," much of it clogging up their on-line systems. If a specific compliance request comes in, a search engine may be used to try and pull out the most likely candidates by matching against appropriate key words.

Now switch that around. If appropriate key words are used on the free-format data when received as part of creating fixed format metadata to accompany the data and you have largely solved the ILM data tiering problem. (This is essentially the approach used by Njini.) Once the metadata is created the software works from the metadata and applies policies or rules to it (and they may update it if a data change occurs). Apart from a speed challenge when the data is first received—it may arrive too fast for real-time metadata creation—this procedure can work. So I wonder why SNIA has not started getting into this.

Fifty companies are already participating in the SNIA initiative and its two associated technical workgroups. These include both application developers from storage vendors and some academic bodies. Among these are some of the "big boys" who are clearly anxious to push the specification. EMC has contributed a C++ with Java Native Interface (JNI) wrapper XAM Library while HP has donated a Java version of the XAM Library. Sun has added code from its Sun StorageTek 5800 (previously "Project HoneyComb") for the Hypertext Transfer Protocol (HTTP) and reference vendor implementation modules (VIMs). This tells me several things:

1. XAM has lift-off and the potential to become the de facto metadata standard for fixed format data. SNIA has the capability and the intention to cultivate a SNIA community for pushing the XAM standard, with an approvals procedure for XAM-compatibility and conformance within software products. It can back this by industry education programmes. That's the good news.
2. There is a danger that, because it is being developed by committee with lots of vested interests, the resulting solution may contain lots of bells and whistles that most do not need and which make it inordinately complicated, slow and unwieldy to use. The best ways of doing things might sometimes be circumvented because one or more of the biggest vendors realise that that approach will undermine their competitive position.
   Storage vendors are first and foremost in the business of making money so the biggest are especially unlikely to support an elegant approach if it cuts them out. Yet such baggage has in the past resulted in standards being ratified, only to be neglected and overtaken by other better approaches.
3. Because of other objectives associated with data management, the primary ILM focus may be lost. There is evidence of this in SNIA's XAM announcement which, by the way, never mentions compliance. SNIA also announced that its Data Management Forum (DMF) is now starting to develop an application-centric standard called a Self-Describing Self-Contained Data Format (SD-SCDF); this, SNIA says, will be coupled with the XAM specification over time. SNIA says: "The SD-SCDF is aimed at providing application developers who adopt XAM, the ability to write a standard, interoperable, long-term preservation format and XAM provides SD-SCDF a strategic catalyst enabling adoption."

Without, admittedly, having investigated the detail, this very description tells me it will introduce a diversion and complexity to what is conceptually a simple enough task. So could XAM end up as a camel (a horse designed by a committee) or perhaps a submerged hippopotamus (a waterhorse designed by several committees)?! That is probably unfair to all the people working hard to produce a good spec covering all eventualities. However, if compliance matters are not central to XAM thinking I am not sure how this horse will be able to stay afloat in practice. I would be more confident if free format content was also being urgently and sensibly addressed within a very short time-frame.

XAM looks interesting and needs to be investigated closely. So I am raising these as my concerns about what will happen to XAM because there is a need and a great opportunity it can address—but I fear this will be missed. My concerns may, of course, be completely unfounded, and I would be delighted to hear from anyone who can put my mind at rest. With the right motivation and full attention to handling free format, XAM could then be of real value in achieving something like full ILM.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Peter Williams (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Michael Warrilow, Director, Hydrasight Posted: 28th March 2008 Copyright Hydrasight © 2008

Hydrasight observes a growing number of enterprise vendors promoting and providing hypervisor technologies, including Microsoft and VMware plus Citrix/XenSource, Oracle and Sun among others. We believe this will result in the increased requirement for heterogeneous hypervisor management, and the potential justification for further investments in IT management software. For the majority of IT organisations however, we predict that the benefits of an infrastructure management ‘suite' that includes heterogeneous hypervisor management are unlikely to justify the required investment before 2011.

While Hydrasight continues to believe that VMware will maintain an unassailable position in server hardware virtualisation for the foreseeable future (refer "HYDRASIGHTS 2008: Server virtualisation"), we predict that multiple hypervisor technologies will increasingly be implemented within enterprises. The key driver for this will be the ‘affinity' of particular hypervisors in particular scenarios. Namely:

• Alignment to hardware vendor (e.g., Sun, IBM)
• Alignment to operating system provider (e.g., Microsoft, Novell, Red Hat)
• Alignment to application vendor (e.g., Oracle)

Hydrasight expects the general conditions and behaviours of enterprise IT to inevitably result in a mixed (hypervisor) environment—as it has with the introduction and adoption of most other technologies. Key contributing factors include varying business unit technology strategies plus ongoing merger and acquisition activity within both organisations and vendors alike.

We believe that key initial requirements for heterogeneous hypervisor management (HHM) will be the ability to ‘migrate' or ‘relocate' virtual machines onto different physical systems plus resilience, discovery/rediscovery and topology mapping. Other initial requirements will be extensions of current needs (e.g., monitoring, provisioning, security and automation). Moving forward, Hydrasight foresees that these requirements will grow to include automated change/configuration management and, ultimately, chargeback.

An unavoidable outcome from the inevitability of multiple hypervisor technologies within enterprise-scale IT environments is a greater rationale for ongoing investment with traditional IT management vendors - e.g., BMC, CA, EMC, HP, IBM and Symantec Corp. We believe that the benefit of being the major incumbent(s), along with ongoing inertia toward migrating IT management environments, which is due to a variety of factors including integration complexity (refer "Peeling back the management integration 'onion'"), will ultimately place the major IT management vendors in a strong position beyond 2011. Hydrasight foresees however that, before 2011, such vendors will continue to lag more-agile IT management ‘start-ups' (e.g., CiRBA, Novell/PlateSpin, Scalent, Surgient).

By 2009/10, as initial requirements for HHM become increasingly commonplace, we believe larger IT management vendors will begin to aggressively acquire smaller HHM vendors in order to fill holes in their portfolio. We therefore advise IT organisations to ensure that investments in HHM functionality are considered tactical, over the medium term (i.e., through 2011), and aligned to/integrated with strategic IT management vendors wherever possible.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Michael Warrilow (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Clay Ryder, President, Sageza Group, Inc. Posted: 10th March 2008 Copyright Sageza Group, Inc. © 2008

The 1394 Trade Association has announced a new specification to quadruple the speed of FireWire to reach 3.2 gigabits per second. The new specification, known as S3200, is backwards-compatible with the IEEE 1394b standard and will be able to use the existing cables and connectors already deployed for FireWire 800 products.

The Silicon Working Group developed the S3200 specification within the 1394 Trade Association, with participation by Symwave, Texas Instruments, LSI Corporation, and Oxford Semiconductor. Since the 1394 arbitration, data and service protocols were not modified for S3200, silicon and software vendors should be able to deploy the faster-speed FireWire quickly. Operating without polling, idle times, or continuous software management, FireWire 800 efficiently delivers more than 97% of its bit rate as payload with FireWire 800 hard drives today moving over 90MBps. S3200 preserves the 1394b design efficiency and is planned to deliver payload speeds reaching 400MBps.

According to the trade association, the best FireWire 800 hard drives move data almost three times as fast as the best hard drives equipped with USB 2.0 and do so with more electrical power to enable operation without an AC adapter and at higher rotational speeds. Alternative cable options that can carry FireWire 100+ meters, even at high speeds, will be available with S3200 and will make this interconnect competitive with eSATA while delivering electrical power, which eSATA does not. Based on the working group's progress, the Trade Association has set a January 2008 date for the specification to enter a ratification process with ratification expected by early February.

Sometimes it is easy to think of interconnects as nothing more than nuts and bolts. Fasteners are generally not all that exciting and can viewed as a commodity. But in the case of FireWire, and its latest high-speed specification, the potential impact is more than a mere commodity could deliver. We have seen some astonishing improvements in copper wire connectivity speeds as of late. Not all that long ago, 100MBps over wire was considered lightning fast, but today 1GBps is met with a yawn of the mundane. The original FireWire was fast for its time, and easily beat early USB implementations. However, neither of these was generally considered for the highest speed interconnects, as this was left to expensive fabrics and optical cabling. Yet today, with this announcement, the notion of 3GBps through a low-cost established interface with considerable backwards capability is now on the cards.

With respect to disk drives, the new S3200 could have substantial impact in the cost of delivering storage both inside and outside of the desktop and the server. Small form factor external drives connected through S3200 offer a single-cable, high-speed, no-power-pack-required storage solution. This simplicity could simplify external disk drive usage in mobile environment where laptops are on the move, but also reduce the dreaded power vampire syndrome for desktop environments where bulky power-sucking transformers pile up around machines, generating heat and sucking power 24x7 even when the disk drive is turned off. Within storage appliances or other internal applications, the simplification of cabling could lead to cost reduction and greater efficiency in solution packaging.

As FireWire has many uses outside disk drives, including camera, cable and satellite set-top boxes, HDTV, and more, the potential for S3200 to compete in high-bandwidth digital media scenarios is considerable. Under S3200, FireWire should offer sufficient throughput to support uncompressed HD signals over longer distances. If further developments permit FireWire to operate over the coaxial cable common in TV installations, the potential of additional data streams coexisting with TV programming should be well received by service providers who are seeking to develop additional revenue streams that leverage the existing infrastructure. But even at present, from the simplistic view of cabling expense, FireWire tends to compare favorably with HDMI, and can easily support storage and other devices that are typically not connective through HDMI. This lends FireWire well to media centers, set-top boxes, etc., as well as its existing base of computing and consumer electronic devices.

Overall, we are intrigued by the potential of S3200 and the potential raising of the bar it implies. As rich media increasingly inundates all aspects of life, the ability to easily and cost-effectively move large numbers of bits between devices becomes all the more important. Further, in an era where green is more than just a color, the leverage of a centralized power distribution source to power devices is well positioned to not only save manufacturing costs and complexity, but lower the overall amount energy consumed by interconnected devices and their users.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Clay Ryder (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Clay Ryder, President, Sageza Group, Inc. Posted: 3rd March 2008 Copyright Sageza Group, Inc. © 2008

EMC has introduced the latest version of EMC Invista version 2.1, its flagship SAN virtualization solution. Invista is an enterprise-class, network-based storage virtualization that combines EMC application software and hardware with intelligent SAN switches from Brocade and Cisco.

The new version of Invista features higher availability for improved data protection, expanded scalability for stronger performance, and enhanced management for better utilization, all of which help organizations keep their infrastructures up and running through both planned and unplanned events. Invista has enhanced high availability through the new distributed control path cluster (CPC) which allows nodes of the CPC to be separated by campus distances, providing for Invista to continue operating in the event of localized failure. The number of virtual volumes and storage elements supported has doubled, and there is a five-fold increase in the number of simultaneous mobility sessions supported. Invista's new heterogeneous pooling and mirroring functionality provides maximum flexibility in creating tiered storage pools and supports mirrored copies across different tiers.

The latest release adds support for IBM DS4000 series arrays, HP PVLinks, and Sun MPxIO path management software and maintains its integration with EMC RecoverPoint to deliver heterogeneous virtual to physical-and virtual to virtual-replication support to enable deployment across multiple sites for disaster tolerance and enhanced availability. In addition, EMC Invista has now been tested, optimized and certified for use with VMware ESX Server 3.0.2, thus enabling organizations to improve their ability to manage, share, and protect the growing amount of information that is being supported by VMware Infrastructure environments. EMC Invista version 2.0 is now available; version 2.1, which includes heterogeneous mirroring and storage pooling, will be available later his month. Invista support for VMware ESX Server 3.0.2 is expected to be posted to the VMware SAN Compatibility Guide before the end of 2007.

It has been about two and a half years, give or take, since EMC originally announced the Invista platform. Given the light speed at which most IT solutions rev their release numbers and ply new features into the marketplace, it is reassuring in many respects that we are now only seeing the second major version of this platform. This is not to say that Invista has been lacking innovation or feature upgrades commensurate with the marketplace, but rather that EMC got the basic principles right the first time, and could take a more measured approach in feature support in favor of allowing end users some time to understand, consume, and realize the full potential of their Invista investment. Nevertheless, there are notable enhancements, specifically the CPC, new heterogeneous pooling and mirroring, and support for additional storage arrays, which we believe add considerably to the overall Invista value proposition. Further, the certification with the latest VMware technology should help assuage any reticence organizations may be having in leveraging the strong operational value proposition that a combined server virtualization and storage virtualization scheme can provide.

Although much of the virtualization discussion to date has focused on servers, the reality is that storage should be treated in most cases in a very similar fashion, as a virtual networked resource. As we have said before, EMC has taken the view that virtualization should occur as far down into the network as possible to take advantage of the increasing intelligence of the underlying network. In this announcement, we see that illustrated in the new support for HP PVLinks and Sun's MPxIO both of which aid in the quest to guarantee multipath I/O connectivity between servers and storage as part of the overall network solution. This also has advantages from a performance as well as a playing-nice-on-the-playground perspective. By preserving the inherent performance (value) that organizations have already deployed in multiple storage arrays and associated hardware and software technologies, EMC is not demanding that customers write off any portion of their past investments in order to gain tangible new value. This remains a pleasant thought to the IT professional who is seemingly all too often cajoled into replacing existing IT investments that still have considerable operational life simply in order to gain a desired or necessary new feature.

Overall, we remain pleased with the underlying vision and direction that Invista has taken. The consistent virtualization and reclamation of underutilized resources story is one that plays well in the acquisition cost- and operational expense-weary IT marketplace of today. Being able to seamlessly move data across multiple storage arrays with differing performance characteristics while maintaining operational integrity is a major must have for most any organization. Solutions such as Invista offer a level of operational efficiency and flexibility that is well positioned to meet this need. Virtualization provides a soothing balm for the IT professional that eases the management of disparate resources under a single view while also allowing greater flexibility in choosing solutions. For vendors, virtualization has been a hot marketing message with commensurate market opportunity, but with marketplace expectation that standards and cooperation trump vendor specific lock in.

To our way of thinking, EMC seems well positioned to continue its SAN virtualization emphasis while furthering the relevance of virtualization across storage, servers, and networking to an ever-larger segment of the IT marketplace.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Clay Ryder (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Clay Ryder, President, Sageza Group, Inc. Posted: 28th February 2008 Copyright Sageza Group, Inc. © 2008

3PAR has announced the introduction of 3cV, a blueprint for the virtual datacenter based upon the combination of the 3PAR InServ Storage Server, HP BladeSystem c-Class, and VMware Infrastructure.

According to the company the modular architectures of the HP BladeSystem c-Class and the 3PAR InServ Storage Server coupled with the increased utilization provided by VMware Infrastructure and 3PAR Thin Provisioning allow organizations to reduce overall storage and server costs by 50% or more. Using VMware VMotion and Distributed Resource Scheduler, HP Virtual Connect and Insight Control, and 3PAR Rapid Provisioning and Dynamic Optimization, customers are able to provision and re-provision physical servers, virtual hosts, and virtual arrays with tailored storage services in a matter of minutes. 3PAR states that its 3cV customers can minimize server floor space through VMware-enabled server consolidation up to 70% with HP BladeSystem density resulting in up to 50% savings. HP Thermal Logic is credited with reduced server power consumption of 30% while the 3PAR InServ Storage Server delivers twice the capacity per floor tile as compared with alternative solutions. In addition, 3PAR thin technologies, Fast RAID 5, and wide striping allow customers to power and cool as much as 75% less disk capacity for a given need. 3PAR InServ Storage Servers and the HP BladeSystem are certified solutions with VMware Infrastructure. 3PAR is a Premier-level member of the VMware Technology Alliance Partner program and HP is a Global-level VMware TAP program member. 3PAR is also a member of the HP BladeSystem Solution Builder Program and works to ensure the BladeSystem c-Class is qualified with 3PAR Utility Storage. The InServ supports remote boot of VMware Infrastructure and the HP BladeSystem. Components of 3cV are purchased separately from their respective providers or from their channel partners.

When it comes to virtualization, we believe standards are essential for the promise of virtualization to come to fruition. Typically, when we see multiple vendors working together to promote solutions that provide a surety of interactivity across the range of IT infrastructure we are heartened and applaud their efforts. In the case of 3PAR, HP, and VMware, these vendors have been focused on virtualization for some time and collectively would have much to offer to their customers. So with this mind, why does this announcement leave us in a less than excited state of mind? The answer is simple: this announcement falls short of the sort of cross-vendor solution that we believe the marketplace is expecting.

In today's lean IT environment, human and capital resources are thinly provisioned (to steal a phrase). For the most part, organizations are looking to the vendor community to help drive standards, provide vision and best practices, and offer solutions that combine technical acumen with the integration and services needed to bring said solutions up and running. This implies a high degree of vendor cooperation and keen preparation of solutions that target specific customer demographics to make the customer's investment as easy as possible to undertake while minimizing the operational and financial exposure to the organization.

In this announcement, 3cV offers a blueprint for datacenter design, some statements of certification between the discrete components, and an invitation for interested parties to buy the solution parts from the various vendors or channel partners. In an age where virtualization, power efficiency, and consolidation are rallying calls to the weary IT professional and CFO, 3cV's response seems pale. As of this writing, a quick search of the HP Website yields no mention of 3cV, and VMware's site only lists a brief mention in a blog that references the 3PAR press release. Absent evidence to the contrary, it would seem that 3PAR's 3cV is a case of preaching to the faithful. If the mentioned partners of this solution cannot muster sufficient excitement to engage at a minimum level of marketing consistency, it is difficult to imagine how 3cV will help drive interest and deployment in this collection of technology in the marketplace. This would not only be a lost opportunity, but would also set the expectation in the eyes of the uninitiated that virtualization is a piecemeal collection of technology, and as such is probably complex, and something that is only relevant/feasible for "the big guys." This would be a real shame.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Clay Ryder (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Peter Williams, Practice Leader - IT Infrastructure Mgmt., Bloor Research Posted: 26th February 2008 Copyright Bloor Research © 2008

As a result of a growing trend towards iSCSI at the expense of fibre channel (FC) connection, together with increasingly high capacity but low cost SATA disks, thin provisioning pioneer 3PAR has spotted a new marketing opportunity—what it calls ‘Nearline for Online' capability.

3PAR has realised that its wide striping architecture can achieve tier 1 disk storage performance using the normally tier 2 iSCSI-connected SATA devices. That means the SATA drives can run critical high performance on-line applications alongside less real-time functions—and save running costs in several ways.

An important factor in making this viable now is that SATA drives, which have been in use for a couple of years, have now demonstrated a high level of reliability whereas initially this was unproven.

Overall data access and transfer performance is not dependent on disk rotation speed only. For instance, in a RAID unit, several platters have their own read-write heads; when the data is ‘wide-striped' across multiple units this provides simultaneous reading of multiple parts of the required data, so ensuring multiplied transfer speeds. 3PAR writes data wide-striped in this way.

3PAR's software is quite happy if there is a mix of 15,000 rpm drives (146GB capacity) or 10,000 rpm drives (400GB capacity) with FC and 750GB enterprise SATA, both within its InServ storage server hardware. However, there is some capacity inefficiency in mixing them in this way.

A major factor in thin provisioning is that the user does not need to plan capacity beyond incremental additions of units as the existing drives become full. Pooling all the capacity within one tier and aggregating it together maximises percentage utilisation and minimises over-provisioning; separately controlling ‘online’ and ‘nearline’ storage is not as efficient or scalable within traditional arrays as both must carry spare capacity. With one disk tier this is eliminated.

Raw performance needs to be proved but 3PAR claims its architecture, with SATA, achieves the performance that most enterprises need—and more reliably than before to better achieve service level guarantees. FC lovers may wish to dismiss the reliability claims for SATA; however, 3PAR's own architecture in any case helps ensure against problems.

3PAR's thin provisioning design uses "chunklets" of data storage which are only allocated when new storage space is needed. If a drive fails, the InServ server needs only to replace these lost chunklets, not the entire failed drive capacity; its ‘Rapid RAID Rebuild' capability picks up the data from associated "raidlets" distributed across drives and builds this into a global spare chunklet pool—also widely spread—so the effect is very fast recovery to a safe state without interruption to work.

This architecture was already present of course; so the announcement is really about the way the market for SATA has been burgeoning. Still, it is worth highlighting since thin provisioning and tiered-storage is about saving costs, and this focus takes it further.

According to Craig Nunes, 3PAR's VP of worldwide marketing, in the last quarter SATA capacity outsold FC at the rate of 65% to 35%; the quarter before was around 35/65—so the trend toward SATA is accelerating. The switch may be less marked in the larger enterprises who have long-established FC and the necessary staff expertise to manage it, but there is still a higher server-side FC cost.

SO, despite 3PAR's enterprise push, the short-term opportunity is potentially even greater among small and medium businesses (SMBs). In general these never have had FC and are now buying iSCSI-SATA in ever larger quantities; but they have capacity and performance issues which are even greater proportionate to business size. "For SMB's why build FC up?" said Nunes; he has a point as long as the need for the highest-performing disks is now obviated.

Here is how the benefits stack up: i) Thin provisioning 3PAR-style achieves considerable overall disk capacity savings over storage virtualised in a standard way (as well as over others' recently bolted-on thin provisioning); ii) the thin provisioning approach also hugely reduces storage capacity planning and management (and therefore the staff costs to do this); iii) because capacity is more predictable, smaller, more incremental, purchases can be made, nearer to when the capacity is needed, a further cost saving; iv) for the disk capacity that is used there is a lower cost with SATA—so the more disk needed the greater the saving; v) there are lower energy costs and heat output. Cost-conscious enterprises should find this proposition attractive.

Of course, saving on disk storage and its running costs is also a green issue. 3PAR is now operating a carbon-neutral environmental policy. It complements disk capacity savings (when a company switches to 3PAR and so saves space), with carbon-offsetting for the rest. The company estimates a saving of equivalent to 12,000 metrics tons of CO2 in 2007 and is going for an even bigger figure this year.

A measure of 3PAR's thin provisioning success to date is perhaps reflected in its recent IPO plus ongoing worldwide expansion. At this week's VMworld Europe event in Cannes, 3PAR is announcing a French storage-focused systems integrator, Antemeta, as its partner covering France and Luxembourg. It is also trumpeting a major hardware/software win with the German Freudenberg Group (Eu5bn turnover, 33,000 employees in 53 countries). Since 3PAR virtualises storage and VMware virtualises servers the two are complementary with little or no overlap.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Peter Williams (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Clay Ryder, President, Sageza Group, Inc. Posted: 14th February 2008 Copyright Sageza Group, Inc. © 2008

Fujitsu Siemens Computers (FSC) has announced a new version of its CentricStor Virtual Tape Appliance. Version 4 features dual-target save, which provides end users the choice of backing up data to either the VTA's disk-based cache or to tape as well as thin provisioning through flexible allocation of storage resources.

The latest version also expands service and maintenance functions, provides the basis for deduplication functionality, and targets the three major pain points that data protection managers face today including continued growth in data volume, increased data retention periods, and the need to support secondary data centers for enhanced disaster tolerance.

Version 4 features a substantial VTL cache (up to 1PB) with support for 1,500,000 logical volumes as well as intelligent data protection whereby end users specify which information should be automatically retained on disk and which should be saved to tape according to desired service level. In addition to existing maintenance services, FSC also offers enhanced proactive services such as Live Monitoring and Health Check which can include FSC monitoring of CentricStor systems with rules-based actions that can be taken automatically in the case of an error along with regular testing and pro active risk assessment. Customers are also offered service management for their complete data protection solution including servers and tape libraries. Pricing information was not disclosed.

The mantra of "tape is dead" continues to be heard throughout the marketplace; however, the reality of simple economics combined with a new "green" twist on tape illustrates how tape continues to enjoy a viable position in many organizations. With all the focus on eco-friendly IT, tape is able to rightly claim that when idle, its storage cartridges consume no energy whereas idle disks (unless powered off) consume some amount of power. While there are many advances in current disk technologies that have substantially reduced power consumption, for archival purposes tape can presently make the greener claim.

Given the astronomical rise in data volumes being stored, this is potentially a non-trivial saving especially for organizations that do not require frequent or instantaneous access to archived data. At the same time, the support for automatic backup of data whether to disk or tape based upon policy addresses a common need in organizations of most any size. This in conjunction with a larger ILM initiative can offer flexibility and overall efficiency gains with respect to data storage.

With this latest release of its CentriStor platform, FSC is demonstrating not only that it sees continued need for virtualized tape solutions, but that the capacity needed is growing as well. 1 petabyte of data is a lot, even if the organization is a Global 1000 enterprise. For the more modest of organizations commonly dubbed as SMBs this degree should meet or exceed their operational storage requirements handily. As such, FSC has a very large addressable market for this offering that, combined with the flexibility in storage approaches offered by CentriStor, places FSC in a desirable, if not enviable, marketplace position.

Useful Links:

• Post Comment | Read Comments
• Send Page Referral
• Contact Clay Ryder (Private)
• Social Bookmarks: Delicious | Digg | Reddit | Facebook | StumbleUpon

By: Peter Williams, Practice Leader - IT Infrastructure Mgmt., Bloor Research Posted: 11th February 2008 Copyright Bloor Research © 2008

Looked at closely, a huge amount of development must have gone into CA Recovery Manager r12, released last week. The new versions of CA ARCserve, CA XOsoft Replication, XOsoft High Availability and XOsoft Assured Recovery that form part of the package represent a major refresh and consolidation of these long-established solutions.

CA asserts that Recovery Manager's objective is to balance data management risks and costs by allowing companies to set the levels of data protection in accordance with whether the information is mission critical (such as in real-time trading), nearly so (for instance internal applications and databases), or less critical, as needed for compliance retention and archiving.

So under the Recovery Manager umbrella, the user can select backup (and recovery), replication with continuous data protection (CDP), automated application failover and even disaster recovery (DR) testing. Recovery Manager controls these functions from one central location. Data can be replicated or moved between sites and locations, and the new CDP provides any point in time recovery.

CA Recovery Manager is a tangible fruit of CA's overarching Enterprise IT Management (EITM) strategy as it brings together previously disparate but related software products under one ‘capability solution’ umbrella. So much is new between these systems that it is difficult to know where to start. (So clearly, those considering upgrade should first ensure the new release carries no serious teething problems.)

Before looking at the detail, my first impression is of a logical, well thought through development, especially attractive to multi-site businesses wanting to control operations centrally.

It will also do much to protect CA's huge user base; for instance, the company boasts 350,000 ARCserve Backup users worldwide. Some of the features, such as CDP and centralised control, have been available for some time from competitors, so this is a needed catch-up; a few other increasingly popular features, like de-duplication, are not yet provided (except through third party appliances). However, there is far more that many potential competitors do not provide—and bringing replication, CDP, DR and backup/restore into the same regime addresses what are historically two parallel markets.

Here, briefly and in overview, is what is under the covers.

CA ARCserve Backup is a platform for long-term data retention and addresses application support. It scales from a single server to a whole enterprise infrastructure but, until r12, had not offered centralised control. Installation is improved and simplified with the help of wizards (as is true for the rest of the suite). Centralisation (which is optional), covers job queues, device management, database management, activity and alerting management and actionable reporting (highlighting processing problems to trigger remedial action).

ARCserve has enhanced encryption capability (including FIPS certified AES 256 bit) and, according to CA, this will not impact performance when being performed disk to disk to tape (D2D2T) or VTL. Other enhancements are new agents to backup MS SharePoint 2007 and multiple VMware sessions respectively; physical to virtual bare metal restores can also be achieved using the VMware agent.

CA is making much also of the new central ARCserve catalogue which uses MS SQL Server (or SQL Express) integrated database; CA says this is more robust, efficient and scalable than its previous database for holding the job and activity log, with lower impact, faster for browsing and searches, and takes less space.

Central licensing management allows licenses to be activated, de-activated and recorded from one management console while, out-of-the-box, ARCserve is now also integrated with CA XOsoft to give remote office and non-disruptive protection.

Finally, it is covered by a new simplified ‘value-based' licensing regime which CA believes is unique; one license will cover all backup requirements and will remain unchanged even if more hardware and/or software is added.

CA XOsoft Replication covers DR through asynchronous but real-time replication over WANs for files and databases while CA XOsoft High Availability has fully automated failover and failback for MS Exchange and SQL, Oracle, IIS and file servers; both these now come with integrated CDP. Again there is centralised and as well as web-based control. A non-disruptive zero-reboot upgrade capability will apply for subsequent releases after r12, so protecting a little against downtime. Synchronisation and replication performance have also been improved.

CA XOsoft Assured Recovery is the non-disruptive remote site automated DR testing capability, and this now works from a separate snapshot of the live data. It is controlled through a script or the ARCserve GUI. This is a powerful element which is probably unique in the industry right now.

The above does little more than scratch the surface of a huge package of upgrades, more attractive because they are logical and fit with the trend to centralised management within larger enterprises. Reports on performance from beta sites are also very good.

From CA's point of view, the integration of ARCserve and XOsoft means both sets of users should see the advantage of upgrading to embrace the extra technology; these users are low-hanging fruit and could alone make CA Recovery Manager r12 a runaway success. Other prospective purchasers may also prefer this route to having to support two vendors' products which may have little commonality.

An i