The next edition of the HP Discover Performance Podcast Series examines how McKesson Corp. accomplished a multi-year, pan-IT management transformation. We’ll learn how McKesson's performance journey, from 2005 to the present, has enabled it to better leverage an agile, hybrid cloud model.
Andy Smith, Vice President of Applications Hosting Services at McKesson, joins host Dana Gardner, Principal Analyst at Interarbor Solutions, to explore how McKesson gained a standardized services orientation to gain agility in deploying its many active applications. [Disclosure: HP is a sponsor of BriefingsDirect podcasts.]
Here are some excerpts:
Gardner: It's hard to believe it's been a full year since we last spoke. What's changed in the last year in how McKesson had been progressing and maturing its applications delivery capabilities?
Smith: Probably one of the things that has changed in the last year is that our performance metrics have continued to improve. We're continuing to see a drop in the number of outages from the standardization and automation. The reliability of the systems has increased, the utilization of the systems has increased, and our system admin ratios have increased. So everything, all the key performance indicators (KPIs), are going in the right direction.
That allowed us to make the next shift, which was to focus on how can we do better at providing capabilities to our customers. How do we do it faster and better through provisioning, because now it's taking less time to do the support side of it.
Gardner: It's really interesting to me that a big part of all this is the provisioning aspect going from fewer manual processes and multiple points of touch to more self-provisioning. How has that worked out?
Smith: It's been very well received. We've been in production now roughly two-and-a-half months. Rather than delivering requests via business requests to add some compute capacity in an average of six months, we’re down to less than four days. I think we can get it down to less than 10 minutes by the time we hit the end of summer.
It's been a challenge to get people to think differently about their processes internal to IT that would allow us to do the automation, but it's been very well received.
Gardner: What were some of the hurdles in terms of trying to get standardized and creating that operating procedure that people could rally behind, self provision, and automate?
Smith: The first piece is just a change in culture. We believe we were customer-centric providers of services. What that really translated to was that we were customer-centric customized providers of services. So every request was a custom request. That resulted in slow delivery, but it also resulted in non-standardized solutions.
One of the most difficult things was getting the architects and engineers to think differently and to understand that standardization would actually be better for the customer. We could get it to them faster, more consistently, and more reliably and, on the back end, provide the support much more cheaply to get that mind shift.
But we were successful. I think everybody still likes to customize, but we haven't had to do that.
Gardner: Just for the edification of our listeners, tell us a bit about McKesson. You’re not just a small mom-and-pop shop.
Smith: No, I think we’re Fortune 14 now, with more than $122 billion in revenue and more than 43,500 employees. We focus specifically on healthcare, how to ensure that whatever is needed by healthcare organizations is there when they need it.
That might be software systems that we write for providers. That could be claims processing that we do for providers. But, the biggest chunk of our business is supply chain, ensuring that the supplies, whether they be medical, surgical, or pharmaceutical, are in the hospital's and providers' hands as soon as they need them.
If a line of business needs to make an improvement in order to capture a need of a customer, with the old way of doing business, it would take me six months to get the computer on the floor. Then they could start their development. Now, you're down to less than a week and days. So they can start their development six months earlier, which really helps us be in a position to capture that new market faster. In turn, this also helps McKesson customers deliver critical healthcare solutions more rapidly to meet today's emerging healthcare needs and enable better health.
Gardner: And there are also some other factors in the market. There's even more talk now about cloud than last year, focusing on hybrid capabilities, where you can pick and choose how to deploy your apps. Then, there's the mobile factor.
Smith: We are recognizing that we have to build that next generation of applications. Part of that is the mobility piece of it, because we have to separate the physical application, the software-as-a-service (SaaS) application from the display device that the customer is going to use. It might be an Android, an iPhone, or something else, a tablet.
So we're recognizing the fact that for next-generation of product, we really have to separate that mobile portion from it, because that display device could be almost anything.
Gardner: We’re here at HP Discover. How have the HP products and services come together to help you not only tackle these technical issues, but to foster the right culture?
Smith: When we talked last year, we had a lot of the support tools in place from HP—operations orchestration, server automation, monitoring tools—but we were using them to do support better. What we're able to do from the provisioning side is leverage that capability and leverage those existing tools.
All we had to do is purchase one additional tool which is a Cloud Service Automation (CSA) that sits on top of our existing tools. So it was a very minor investment, and we were able to leverage all the support tools to do the provisioning side of the business. It was very practical for us and relatively quick.
Gardner: Of course, a big emphasis here at HP Discover is HP Converged Cloud and talking about these different hybrid models. How has the automation provisioning services orientation and standardization put you in a place to be able to avail yourselves of some of these hybrid models and the efficiencies and speed that come with that? How do they tie together—what you’ve done with applications now and what you can perhaps do with cloud?
Smith: We’ll be the first to admit that providing the services internally is not necessarily always the best. We may not be the cheapest and we may not be the most capable. By getting better at how we do provisioning and how we do our own internal cloud frees up resources, and those resources now can start thinking about how we work with an external provider.
That's a lot of concern for us right now, because there is that risk factor. Do you put your intellectual property (IP) out there? Do you put your patients’ medical records out there? How do you protect it? And so there are a lot of business rules and contracting issues that we have to get through.
From a technology standpoint, we know we can do it. We’ve done it in the labs. We’ve provisioned out to third-party providers. It all works from a technology standpoint with the tools we have. Now we have to get through the business issues.
It's fortunate, in some ways, that HP is on the same journey. We partner on a lot of these things. When we brought CSA in, it was one of the earlier releases, and now we’ve partnered with them through the Customer Advisory Boards (CABs) and other methods. They continue to enhance this to meet our needs, but also to meet their needs.
Gardner: Now that you've been on this journey from 2005, where do you see yourselves in a couple of years?
Smith: Because we’re in healthcare, very similar to banking, we've hit a point where we don't believe we can afford to be down anymore.
Instead of talking about three nines, four nines, or five nines, we're starting to talk about, how we ensure the machines are never down, even for planned maintenance. That's taking a different kind of infrastructure, but that’s also taking a different kind of application that can tolerate machines being taken offline, but continue to run.
That's where our eye is, trying to figure out how to change the environment to be constantly on.
If the application isn't smart enough to tolerate a piece of machine going down, then you have to redesign the application architecture. Our applications are going to have to scale out horizontally across the equipment as the peaks and valleys of the customer demands change through the day or through the week.
The current architecture doesn't scale horizontally. It scales up and down. So you end up with a really big box that’s not needed some times of the day. It would be better if we could spread the load out horizontally.
Gardner: So, just to close out, we have to think about applications now in the context of where they are deployed, in a cloud spectrum or continuum of hybrid types of models. We also have to think about them being delivered out to a variety of different endpoints.
What do you think you’ll need to be doing differently from an application development, deployment, and standardization perspective in order to accomplish both that ability to deploy anywhere and be high performance, as well as also be out on a variety of different end points?
Smith: The reality is that part of our journey over the last several years has been to consolidate the environment, consolidate the data centers, and consolidate and virtualize the servers. That's been great from a customer cost standpoint and standardization standpoint.
But now, when you're starting to deliver that SaaS mobile kind of application, speed of response to the customer, the keystroke, the screen refresh, are really important. You can't do that from a central data center. You've got to be able to push some of the applications and data out to regional locations. We’re not going to build those regional locations. It's just not practical.
That's where we see bringing in these hybrid clouds. We’ll host the primary app, let's say, back in our corporate data center, but then the mobile piece, the customer experience piece, is going to be have to be hosted in data centers that are scattered throughout the country and are physically much closer to where the customer is.
Gardner: Of course, that’s going to require a different level of performance monitoring and management.
Smith: Exactly, because then you really have to monitor the application, not just the server at the back-end. You’ve got to be watching that performance to know whether you have a local ISP that’s come down, if you have got a local cloud that’s come down. You’re going to really have to be watching the endpoints so you can see that customer experience. So it is a different kind of application monitoring.