The ITKO Blog: Service Virtualization and Application Lifecycle Optimization - CA Technologies

Modernizing Legacy Apps: Still Sexy?

Yes, legacy modernization might be about as un-sexy as you can get as far as IT practices go. But we know that these venerable IT assets are a reality of any serious operating enterprise, and making those mainframes and core systems work in today's architecture is still a very hot topic for discussion.

Our John Michelsen recently joined in a good discussion on the eBizQ Forum. Question posed by one of our favorite bloggers, Joe McKendrick: Are Web services protocols such as SOAP and REST and AJAX effective for building SOA off of mainframe, large systems or legacy environments?  Or is message-oriented middleware more advisable?

Well, yes, and yes as John mentions... they better be effective at bringing those systems into a modern architecture, if you are using them. Many developers and architects can get quite dogmatic about SOA vs. REST, or how MOM solutions like ESBs are good or evil.  It sounds like a blanket statement, but the longer we are in the business of aligning IT around business, the more ambivalent we must become about these distinctions.

Anyway I think the Q&As going on there are worth reading, and we are seeing a lot of the "usual suspects" in SOA punditry and integration expertise weighing in.

Keeping Development Close in the Cloud with ZDNet's Gardner

Continuing our discussion on "Personal Cloud" technology from last week, I had a chance to catch up with ZDNet/BriefingsDirect analyst Dana Gardner, whom we also ran into at the most recent HP Software Universe event in Las Vegas.

In his latest blog: "The Cloud Gets Up Close and Personal," Dana was particularly interested in virtualization and Cloud Computing in terms of freeing up developers from dependencies and costs by replicating the target environment on their local workstation, a capability which is entirely possible today:

"The confluence of SaaS and cloud with application development and the test phase is changing rapidly..." he observed. "Compressing the test phase into the development and production becomes more feasible. And as virtualization becomes more common, building an application or service in its own runtime stack bubble from inception to sunset starts to make sense..."

Indeed, we feel that ANYTHING you can do to eliminate waiting on incomplete environments, unavailable systems, unexpected costs and fees, all of those things are what we need to solve if we are going to really see the value we expect from today's distributed, flexible kinds of apps -- which will likely lean heavily on the Cloud as a deployment strategy.

Having a personal-size development environment is almost as popular with developers as say, a personal-size pizza. And in terms of the costs IT departments currently must spend to enable robust test environments, the cost to the business can now seem as small as adding one topping to that personal pizza. He quoted me from the briefing:

"First, the developer working on a cloud application is free to work anywhere, anytime regardless of whether the real cloud services are available or accessible. If a cloud service for a shopping cart is down for some reason, developers are not impacted since their version of the service is on their laptop. They can also code when they are on a plane, or in another environment with no access to the cloud.

Second, although this is probably first in the minds of budget conscious IT managers, the developer is not running up charges for accessing the cloud services..." [Chris Kraus, iTKO]

Now, this sounds great. But what are the unaddressed frontiers of such an approach? Well, for one thing, Dana is looking for the whole package coming out of it:

"...while we’re combining all the elements of an application and platform from cradle to grave, why not tune the whole package before, during and after development too … then load the entire package as a portable cloud-supported production unit?

Now, that’s a “personal” cloud (I prefer cloud service nodule), but with high service performance output, and far less time in cost in the total lifecycle. Higher overall quality too. What do you think?"

We like it - however with critical deployed systems, even if they are very well tested before they are packaged and deployed by the developer, you still MUST test continuously in deployment, as all of the components and usage patterns the developer needed to have stable on the desktop, may not be so stable or used in known ways in Cloud deployment. We can get most of the way there, but probably not 100% there, as much as I'd like to be proven wrong someday.

In addition we will still see these efforts put into Public and Private Cloud environments, due to the ease of administration and economy of scale benefits of provisioning shared "anytime/anywhere" environments for the software lifecycle.

I do believe we will continue to see innovation in this realm - the personal cloud or "local virtualized" environment enabling multiple individuals and teams to work in parallel - married to next-generation application packaging and deployment options. Indeed there are already some nice technologies out there, which we will discuss with partners in future posts.

Could Open Cloud Computing Continue to Coalesce?

I wrote recently about the public sector (US Federal Government) in playing a leadership role in cloud computing. The private sector is also making moves, as James Urquhar wrote in IBM, Microsoft, others align on open clouds.

James noted that representatives of Cloud Computing Interoperability Forum (CCIF), CloudCamp, Cisco, IBM, Intel, Microsoft, and the IEEE-ISTO met. They agreed on a shared goal to promote use and awareness of open and interoperable cloud computing. They also discussed the ability to enable participants to be able to contribute and use results of broad community collaboration. There was also discussion of the possibility of a trade association or marketing association for cloud computing, but no actions were agreed upon at that time.

These kinds of activities typical for a technology that is still developing. I agree with the hope James expressed for open cloud standards operating in a cooperative, rather than competitive, fashion. I hope vendors are able to move forward with some vision around shared interests for Public clouds, while at the same time we see companies employing more specialized and secure "Private Clouds" and other hybrids such as the "Personal Cloud" we recently talked about, for specific software lifecycle purposes.

The release of an Open Cloud Manifesto, originally authored by IBM, was not without some disagreement, as Microsoft objected, and there were plenty of differing opinions thrown around the blogs I read after this announcement. The group's site shows that many players have signed up but, as James notes in another post, the "big four" of cloud computing, Amazon.com, Microsoft, Google and Salesforce.com, are not signatories so far. You can find the Open Cloud Manifesto at opencloudmanifesto.org.

It notes five challenges to adoption: security, data and application interoperability, data and application portability, governance and maintenance, and metering and monitoring. Several of the challenges this collective is concerned with, are also factors that in our experience, customers are very concerned with; especially the quality and validation aspects of governance, and the portability of Cloud services through virtualization of software components hosted in the cloud to eliminate some dependencies and per-use costs.

The manifesto concludes that it is meant to "begin the conversation, not define it." Obviously these kinds of industry groups are promoted merely as representatives of common interests, and not legal entities. However, we look forward to the rest of the story as the cloud coalesces.

On-Demand App Dev Strategies with Compuworld's Mark Hall

I recently had a chance to catch up with Mark W. Hall, an analyst and writer for Compuworld. Mark was especially interested in the concept of bringing development and testing environments into readiness in his blog "On Demand App/Dev? Not yet, but closer."

Mark talked to several vendors that are in application quality related fields, as this can often be an underdeveloped aspect of development. In other words, we are now take an extremely "agile" approach to developing new functionality, and the quality or ability to test all of the components and interconnectedness in an application's environment just hasn't kept up.

Here's an excerpt of Mark's article:

"At iTKO Inc. in Dallas the company's LISA product lets you tests services (in the cloud or in your data center) virtually. Chief Geek John Michelsen says LISA watches how your app interacts with a given service over time, then it creates a virtual instance of the service with all of its behavioral quirks and features that you can test as often as you like. Michelsen claims it is unrealistic for a developer to test against a live service."

[Just as a clarification - I believe we definitely must enable developers to test live services for checking implementation - however I do believe developers and testers cannot get access to the production environment to do so 95% of the time they really need it, especially if it is already handling live customer transactions...]

Mark also asked about hosting these virtual environments in the Cloud, and a resounding YES, we already see Cloud providers and partners offering up Virtual Services capabilities, playing nicely alongside other virtual servers and machines that make up a well-rounded test environment. Look for more from us on this topic soon.

eBizQ discussion: Are we Meaningfully Applying SOA Governance?

Editor of eBizQ Peter Schooff recently asked a small group of us pundits to post questions to their new expert Forum site, and this week he ran a question from me that we've been asking our customers: Are we meaningfully Applying SOA Governance?

"While we've seen a large number of firms realizing significant value from SOA approaches, are companies truly leveraging SOA Governance principles across their entire design, development and delivery lifecycle? Are solutions such as Service Management, UDDI Registry/Repositories, Testing and Validation being employed as part of this effort? How can SOA Governance accomplish greater adoption levels in today's enterprise environments?"

Our friend Miko Matsumura from Software AG said a resounding yes, and chalks success up to behavioral change and not tooling: "Today, the 'install software and it just works' concept is all but moot. Despite the desire of software that 'just works', the value obtained by software is pretty much only attainable when the organization changes its behavior, for example in a business process."

Author Michael Poulin chimed in and extended the business responsibility aspect, while cautioning that any organization leveraging the tools before the principles has already lost. Fair point, and one reinforced by Joe McKendrick later in the column.

JP Morgenthal also noted from a technology perspective: "I believe a lot of implementations give 'lip service"' to governance and will implement the registry as part of the ESB support requirements, but do not have a plan for how to move beyond this phase of implementation." This is very much in line with what we've seen. Perhaps this will be the year people start utilizing these tools hand in hand.

This particular topic has been ongoing for some time here (see our blog post from a couple years ago on "Wringing Value out of UDDI Enforcement and Policy" or click on our SOA Governance category). We often find that without disciplined practices, and without a proper means of validation and enforcement of the Policies defined in a Registry/Repository or Service Management solution, the SOA Governance initiative faces some challenges.

In any case I am really enjoying the caliber of these ongoing discussions, and look for us to comment more when interesting topics come up.

Who needs a "Personal Cloud?" - Gualtieri

With all of the hulabaloo around Cloud Computing, you'd think the LAST thing we need is yet another kind of cloud, in addition to Public Clouds and Private Clouds (secured within a company's domain). Forrester's Mike Gualtieri, an application development analyst we have enjoyed talking to in the past, agrees in his blog "Cloudmania: Developers Need A Personal Cloud" that we don't need a new cloud for everything in principle, while proposing another cloud anyway.

"What developers could use is a Personal Cloud that would allow them to configure their local environment in multiple way and take it with them wherever they go. I know this sounds like virtualization and it is to some extent, but extend PC virtualization with cloud concepts and you get the Personal Cloud."

What is a "personal cloud" anyway? Is it the "Most Private Cloud?" In essence it is a way for developers to completely isolate themselves from dependencies, by having all of the services and resources they need to work and test in their own "personal cloud" environment right on their own computer. No connection to the Internet or a hosted service needed.

At this point you may ask yourself if this can even be called a Cloud, and if it is even relevant to the real world, much less plausible. Indeed, many of the comments on Mike's blog are challenging the feasibility of a "Personal Cloud" altogether. If you are talking about working with enterprise applications, you can't possibly replicate monolithic mainframes, systems of record, and simulate complex transaction and service environments right on the developer's laptop. Right?

In reality this kind of thing is happening today with the practice of Service Virtualization (you can read a paper on the concept here), when coupled with a localized Virtual Environment (or virtual "appliance") that allows a very genuine simulation of the developers' dependencies to run locally. Even if the moniker "Personal Cloud" doesn't stick, we think he is onto something.

Mike mentions Moore's law: exponentially increased processing power and capacity on the developer's desktop at lower prices. That makes what seemed to be an impossibility 5-10 years ago, a very real proposition today. As developers have more local computing power at the ready, they will be increasingly able to model sophisticated test and development environments within a VM, provided they are detailed, robust, and can perform close enough to reality to be relevant.

Anyway a good original blog and this idea is worth watching as it develops. While there are compelling reasons why Public and Private Cloud deployment are here to stay, don't be surprised when developers ask for the ability to truly work "off the grid" sometimes to move their lifecycles into parallel and eliminate the costs and constraints of access to the environment.