Enterprise -> Technology
By: Dana Gardner, Principal Analyst, Interarbor Solutions
Published: 3rd July 2014
Copyright Interarbor Solutions © 2014
The advent of the application programming interface (API) economy has forced a huge, pressing need for organizations to both seek openness and improve security for accessing mobile applications, data, and services anytime, anywhere, and from any device.
Awash in inadequate passwords and battling subsequent security breaches, business and end-users alike are calling for improved identity management and federation technologies. They want workable standards to better chart the waters of identity management and federation, while preserving the need for enterprise-caliber risk remediation and security.
Meanwhile, the mobile tier is becoming an integration point for scads of cloud services and APIs, yet unauthorized access to data remains common. Mobile applications are not yet fully secure, and identity control that meets audit requirements is hard to come by. And so developers are scrambling to find the platforms and tools to help them manage identity and security, too.
Clearly, the game has changed for creating new and attractive mobile processes, yet the same old requirements remain wanting around security, management, interoperability, and openness.
BriefingsDirect assembled a panel of experts to explore how to fix these pressing needs: Bradford Stephens, the Developer and Platforms Evangelist in the CTO’s Office at Ping Identity; Ross Garrett, Senior Director of Product Marketing at Axway, Kelly Grizzle, Principal Software Engineer at SailPoint Technologies. Welcome, Kelly. The sponsored panel discussion is moderated by me, Dana Gardner, Principal Analyst at Interarbor Solutions.
Here are some excerpts:
Gardner: We are approaching the Cloud Identity Summit 2014 (CIS), which is coming up on July 19 in Monterey, Calif. There's a lot of frustration with identity services that meet the needs of developers and enterprise operators as well. So let’s talk a little bit about what’s going on with APIs and identity.
What are the trends in the market that keep this problem pressing? Why is it so difficult to solve?
Stephens: Well, as soon as we've settled on a standard, the way we interact with computers changes. It wasn’t that long ago that if you had Active Directory and SAML and you hand-wrote security endpoints of model security products, you were pretty much covered.
But in the last three or four years, we've gone to a world where mobile is more important than web. Distributed systems are more important than big iron. And we communicate with APIs instead of channels and SDKs, and that requires a whole new way of thinking about the problem.
Garrett: Ultimately, APIs are becoming the communication framework, the fabric, in which all of the products that we touch today talk to each other. That, by extension, provides a new identity challenge. That’s a lot of reason why we've seen some friction and schizophrenia around the types of identity technologies that are available to us.
So we see waves of different technologies come and go, depending on what is the flavor of the month. That has caused some frustration for developers, and will definitely come up during our Cloud Identity Summit in a couple of weeks.
Grizzle: APIs are becoming exponentially more important in the identity world now. As Bradford alluded to, the landscape is changing. There are mobile devices as well as software-as-a-service (SaaS) providers out there who are popping up new services all the time. The common thread between all of them is the need to be able to manage identities. They need to be able to manage the security within their system. It makes total sense to have a common way to do this.
APIs are key for all the different devices and ways that we connect to these service providers. Becoming standards based is extremely important, just to be able to keep up with the adoption of all these new service providers coming on board.
Gardner: As we describe this as the API economy, I suppose it’s just as much a marketplace and therefore, as we have seen in other markets, people strive for predominance. There's jockeying going on. Bradford, is this a matter of an architectural shift? Is this a matter of standards? Or is this a matter of de-facto standards? Or perhaps all of the above?
Stephens: It’s getting complex quickly. I think we're settling on standards, like it or not, mostly positively. I see most people settling on at least OAuth 2.0 as a standard token, and OpenID Connect for implementation and authentication of information, but I think that’s about as far as we get.
There's a lot of struggle with established vendors vying to implement these protocols. They try to bridge the gap between the old world of say SAML and Active Directory and all that, and the new world of SCIM, OAuth, OpenID Connect. The standards are pretty settled, at least for the next two years, but the tools, how we implement them, and how much work it takes developers to implement them, are going to change a lot, and hopefully for the better.
Garrett: We have identified a number of new standards that are bridging this new world of API-oriented connectivity. Learning from the past of SAML and legacy, single sign-on infrastructure, we definitely need some good technology choices.
The standards seem to be leading the way. But by the same token, we should keep a close eye on the market changing with regards to how fast standards are changing. We've all seen things like OAuth progress slower than some of the implementations out there. This means the ratification of the standard was happening after many providers had actually implemented it. It's the same for OpenID Connect.
We are in line there, but the actual standardization process doesn’t always keep up with where the market wants to be.
Gardner: We've been this play out before that the standards can lag. Getting consensus, developing the documentation and details, and getting committees to sign off can take time, and markets move at their own velocity. Many times in the past, organizations have hedged their bets by adopting multiple standards or tracking multiple ways of doing things, which requires federation and integration.
Kelly, are there big tradeoffs with standards and APIs? How do we mitigate the risk and protect ourselves by both adhering to standards, but also being agile in the market?
Grizzle: That’s kind of tricky. You're right in that standards tend to lag. That’s just part and parcel of the standardization process. It’s like trying to pass a bill through Congress. It can go slow.
Something that we've seen some of these standards do right, from OAuth and from the SCIM perspective, is that both of those have started their early work with a very loose standardization process, going through not one of the big standards bodies, but something that can be a little bit more nimble. That’s how the SCIM 1.0 and 1.1 specs came out, and they came out in a reasonable time frame to get people moving on it.
Now that things have moved to the Internet Engineering Task Force (IETF), development has slowed down a little bit, but people have something to work with and are able to keep up with the changes going on there.
I don’t know that people necessarily need to adopt multiple standards to hedge their bets, but by taking what’s already there and keeping a pulse on the things that are going to change, as well as the standard being forward-thinking enough to allow some extensibility within it, service providers and clients, in the long run, are going to be in a pretty good spot.
Gardner: We've talked a few technical terms so far, and just for the benefit of our audience, I'd like to do a quick primer, perhaps with you Bradford. To start: OAuth, this is with the IETF now. Could you just quickly tell the audience what OAuth is, what it’s doing, and why it’s important when we talk about API, security and mobile?
Stephens: OAuth is the foundation protocol for authorization when it comes to APIs for web applications. OAuth 2 is much more flexible than OAuth 1.
Basically, it allows applications to ask for access to stuff. It seems very vague, but it’s really powerful once you start getting the right tokens for your workflows. And it provides the same foundation for everything else we do for identity and APIs.
The best example I can think of is when you log into Facebook, and Facebook asks whether you really want this app to see your birthday, all your friends’ information, and everything else. Being able to communicate all that over the OAuth 2.0 is a lot easier than how it was with OAuth 1.0 a few years ago.
Gardner: How about OpenID Connect. This is with the OpenID Foundation. How does that relate, and what is it?
When you access an API and you authenticate, you choose a scope, and one of the most common scopes is OpenID Profile. This OpenID Profile will just have things like your username, maybe your address, various other pieces of identity information, and it describes who the "you" is, who you are.
Gardner: And SCIM, you mentioned that Kelly, and I know you have been involved with it. So why don’t you take the primer for SCIM, and I believe it’s Simple Cloud Identity Management?
Grizzle: That's the historical name for it, Simple Cloud Identity Management. When we took the standard to the IETF, we realized that the problems that we were solving were a little bit broader than just the cloud and within the cloud. So the acronym now stands for the System for Cross-domain Identity Management.
That’s kind of a mouthful, but the concept is pretty simple. SCIM is really just an API and a schema that allows you to manage identities and identity-related information. And by manage them, I mean to create identities in systems to delete them, update them, change the entitlements and the group memberships, and things like that.
Gardner: From your perspective, Kelly, what is the relationship then between OAuth and SCIM?
Grizzle: OAuth, as Bradford mentioned, is primarily geared toward authorization, and answers the question, "Can Bob access this top-secret document?" SCIM is really not in the authorization and authentication business at all. SCIM is about managing identities.
OAuth assumes that an identity is already present. SCIM is able to create that identity. You can create the user "Bob." You can say that Bob should not have access to that top-secret document. Then, if you catch Bob doing some illicit activity, you can quickly disable his account through a SCIM call. So they fit together very nicely.
Gardner: In the real world, developers like to be able to use APIs, but they might not be familiar with all the details that we've just gone through on some of these standards and security approaches.
How do we make this palatable to developers? How do we make this something that they can implement without necessarily getting into the nitty-gritty? Are there some approaches to making this a bit easier to consume as a developer?
Stephens: As a developer who's relatively new to this field—I worked in database for three years—I've had personal experience of how hard it is to wrap your head around all the standards and all these flows and stuff. The best thing we can do is have tool providers give them tools in their native language, or in the way developers work with things.
This needs well-documented, interactive APIs—things like Swagger—and lots of real-world code examples. Once you've actually done the process of authentication through OAuth, getting a JSON Web Token, and getting an OpenID Connect profile, it’s really simple to see how it all works together, if you do it all through a SaaS platform that handles all the nitty-gritty, like user creation and all that.
If you have to roll your own, though, there's not a lot of information out there besides the WhitePages and Wall Post. It’s just a nightmare. I tried to roll my own. You should never roll your own.
So having SaaS platforms to do all this stuff, instead of having documents, means that developers can focus on providing their applications, and just understand that I have this media and project, not about which tokens carry information that accesses OAuth and OpenID Connect.
I don’t really care how it all works together; I just know that I have this token and it has the information I need. And it’s really liberating, once you finally get there.
So I guess the best thing we can do is provide really great tools that solve the identity-management problems.
Garrett: Tools, that’s the key point here. Whether we like it or not, developers tend to be kind of lazy sometimes and they certainly don’t have the time or the energy to understand every facet of the OAuth specification. So providing tools that can wrap that up and make it as easy to implement as possible is really the only way that we get to really secure mobile applications or any API interaction. Because without a deep understanding of how this stuff works, you can make pretty fundamental errors.
Having said that, at least we've started to take steps in the right direction with the standards. OAuth is built at least with the idea of mobile access in mind. It’s leveraging REST and JSON types, rather than SOAP and XML types, which are really way too heavyweight for mobile applications.
So the standards, in their own right, have taken us in the right direction, but we absolutely need tools to make it easy for developers.
Grizzle: Tools are of the utmost importance, and some of the identity providers and people with skin in the game, so to speak, are helping to create these tools and to open-source them, so that they can be used by other people.
Another thing that Ross touched on was keeping the simplicity in the spec. These things that we're addressing—authorization, authentication, and managing identities—are not extremely simple concepts always. So in the standards that are being created, finding the right balance of complexity versus completeness and flexibility is a tough line to walk.
With SCIM, as you said, the first initial of the acronym used to stand for Simple. It’s still a guiding principle that we use to try to keep these interactions as simple as possible. SCIM uses REST and JSON, just like some of these other standards. Developers are familiar with that. Putting the burden on the right parties for implementation is very important, too. To make it easy on clients, the ones who are going to be implementing these a lot, is pretty important.
Gardner: Do these standards do more than help the API economy settle out and mature? Cloud providers or SaaS providers want to provide APIs and they want the mobile apps to consume them. By the same token, the enterprises want to share data and want data to get out to those mobile tiers. So is there a data-management or brokering benefit that goes along with this? Are we killing multiple birds with one set of standards?
Garrett: The real issue here, when we think about the new types of products and services that the API economy is helping us deliver, is around privacy and ultimately customer confidence. Putting the user in control of who gets to access which parts of my identity profile, or how contextual information about me can perhaps make identity decisions easier, allows us to lock down, or better understand, these privacy concerns that the world has.
Identity isn’t the most glamorous thing to talk about—except when it all goes wrong and some huge leak makes the news headlines, or some other security breach has lost credit-card numbers or people’s usernames and passwords.
In terms of how identity services are developing the API economy, the two things go hand in hand. Unless people are absolutely certain about how their information is being used, they simply choose not to use these services. That’s where all the work that the API management vendors and the identity management vendors are doing and bringing that together is so important.
Gardner: You mentioned that identity might not be sexy or top of mind, but how else can you manage all these variables on an automated or policy-driven basis? When we move to the mobile tier, we're dealing with multiple networks. We're dealing with multiple services ... cloud, SaaS, and APIs. And then we're linking this back to enterprise applications. How other than identity can this possibly be managed?
Stephens: Identity is often thought of as usernames and passwords, but it’s evolving really quickly to be so much more. This is something I harp on a lot, but it’s really quickly becoming that who we are online is more important than who we are in real life. How I identify myself online is more important than the driver's license I carry in my wallet.
As you know, your driver’s license is like a real-life token of information that describes what you're allowed to do in your life. That’s part of your identity. Anybody who has lost their license knows that, without that, there's not a whole lot you can do.
Bringing that analogy back to the Internet, what you're able to access and what access you're able to give other people or other applications to change important things, like your Facebook posts, your tweets, or go through your email and help categorize that is important. All these little tasks that help define how you live, are all part of your identity. And it’s important that developers understand that because any connected application is going to have to have a deep sense of identity.
Gardner: Let me pose the same question, but in a different way. When you do this well, when you can manage identity, when you can take advantage of these new standards that extend into mobile requirements and architectures, with the API economy in mind, what do you get? What does it endow you with? What can you do that perhaps you couldn’t do if you were stuck in some older architectures or thinking?
Grizzle: Identity is key to everything we do. Like Bradford was just saying, the things that you do online are built on the trust that you have with who is doing them. There are very few services out there where you want completely anonymous access. Almost every service that you use is tied to an identity.
So it’s of paramount importance to get a common language between these. If we don’t move to standards here, it's just going to be a major cost problem, because there are a ton of different providers and clients out there.
If every provider tries to roll their own identity infrastructure, without relying on standards, then, as a client, if I need to talk to two different identity providers, I need to write to two different APIs. It’s just an explosive problem, with the amount that everything is connected these days.
So it’s key. I can’t see how the system will stand up and move forward efficiently without these common pieces in place.
Gardner: Do we have any examples along these same lines of what do you get when you do this well and appropriately based on what you all think is the right approach and direction? We've been talking at a fairly abstract level, but it really helps solidify people’s thinking and understanding when they can look at a use-case, a named situation or an application.
Stephens: If you want a good example of how OAuth delegation works, building a Facebook app or just working on Facebook app documentation is pretty straightforward. It gives you a good idea of what it means to delegate certain authorization.
Likewise, Google is very good. It’s very integrated with OAuth and OpenID Connect when it comes to building things on Google App Engine.
So if you want to secure an API that you built using Google Cloud on Google App Engine, which is trivial to do, Google Cloud Endpoints provides a really good example. In fact, there is a button that you can hit in their example button called Use OAuth and that OAuth transports OpenID Connect profile, and that’s a pretty easy way to go about it.
Garrett: I'll just take a simple consumer example, and we've touched on this already. It was the idea in the past, where every individual service or product is offering only their identity solution. So I have to create a new identity profile for every product or service that I'm using. This has been the case for a long time in the consumer web and in the enterprise setting as well.
So we have to be able to solve that problem and offer a way to reuse existing identities. It involves so taking technologies like OpenID Connect, which is totally hidden to the end user really, but simply saying that you can use this existing identity, your LinkedIn or Facebook credentials, etc., to access some new products, takes a lot of burden away from the consumer. Ultimately, that provides us a better security model end to end.
The thing that these new identity service providers have been offering has, behind the scenes, been making your lives more secure. Even though some people might shy away from using their Facebook identity across multiple applications, in many ways it’s actually better to, because that’s really one centralized place where I can actually see, audit, and adjust the way that I'm presenting my identity to other people.
That’s a really great example of how these new technologies are changing the way we interact with products everyday.
Grizzle: At SailPoint, the company that I work for, we have a client, a large chip maker, who has seen the identity problem and really been bitten by it within their enterprise. They have somewhere around 3,500 systems that have to be able to talk to each other, exchange identity data, and things like that.
The issue is that every time they acquire a new company or bring a new group into the fold, that company has its own set of systems that speak their own language, and it takes forever to get them integrated into their IT organization there.
So they've said that they're not going to support every app that these people bring into the IT infrastructure. They're going with SCIM and they are saying that all these apps that come in, if they speak SCIM, then they'll take ownership of those and pull them into their environment. It should just plug in nice and easy. They're doing it just because of a resourcing perspective. They can't keep up with the amount of change to their IT infrastructure and keep everything automated.
Gardner: I want to quickly look at the Cloud Identity Summit that’s coming up. It sounds like a lot of these issues are going to be top of mind there. We're going to hear a lot of back and forth and progress made.
Does this strike you, Bradford, as a tipping point of some sort, that this event will really start to solidify thinking and get people motivated? How do you view the impact of this summit on cloud identity?
Stephens: At CIS, we're going to see a lot of talk about real-world implementation of these standards. In fact, I'm running the Enterprise API track and I'll be giving a talk on end-to-end authentication using JAuth, OAuth, and OpenID Connect. This year is the year that we show that it's possible. Next year, we'll be hearing a lot more about people using it in production.
All fields must be completed to submit a comment. Email addresses are passed through to the author so they can contact you directly if needed.
Published by: electronicdawn Ltd.