Why Business Transaction Management Is Hot in 2009

Business Transaction Management (BTM) has been getting a lot of attention lately: IT organizations want it, and every major vendor in systems and performance management has rushed to claim they do it.

The promise of BTM is to finally be able to manage IT from a business perspective and make sense of the complexity of modern applications broken down in services that run across environments (real and virtual) distributed behind the firewall and in the cloud.

Business managers want IT to talk in a language they can understand, and they want applications to be reliable. IT executives want to get out of reactive mode and be a partner in creating value for the business. The people in IT operations who work tirelessly to keep things running want to catch a problem before it becomes one — and when a problem does arise, they want to be able to pinpoint it and fix it quickly. Application developers want to get out of the fingerpointing game of war rooms and focus on new capabilities … and so on.

Moreover, business transactions touch many layers of infrastructure and applications across wide networks, with downtime and poor performance directly impacting revenue, brand and customer loyalty/retention. Thus, the need for business technology services that can ensure highly reliable transactions and end-user performance via rapid problem diagnosis and resolution — while at the same time support change and capacity planning that ultimately delivers better service — is paramount. Can BTM live up to these expectations?

From Chasing Fires to Chasing Value

CIOs strive to play a strategic role in the business: driving the transformations that produce top and bottom line value. To do so, they need to move past just keeping the lights on, but it is hard to be proactive when you are overwhelmed chasing fires and trying to fix problems.

(click on image to enlarge)

How can IT grow from chasing fires to chasing value and enable the CIO to become more strategic?

First, IT needs to operate and communicate in a business context, not just in a technology context. Second, IT management functions need to evolve from a component focus (end-users, applications and infrastructure) to a business transaction focus, so solving problems and planning is done efficiently and holistically.

While these two points seem obvious, actually making them happen in a mission-critical, distributed, service-oriented environment that is constantly changing is quite tricky, but it can pay huge dividends.

A Green Light Is a Green Light – or Is It?

Having all-system-lights-green from an IT operations perspective is useless if an online banking customer cannot pay a bill. Deprovisioning a low-utilization server will be a disaster if you don’t know that your insurance agents’ claims processing touches it. Rolling out a new service will be a failure and infuriate customers if it will compete for resources with other applications and cause unacceptable performance.

Why are these types of problems plaguing organizations? Why is it so difficult to manage the technologies that power business transactions and interactions?

Complexity!

It used to be fairly simple to manage data centers in a world of monolithic applications and systems. But then systems became distributed; infrastructure was scattered across n-tiers and is now getting virtualized; IT services have been outsourced; applications are broken down into components and Web services.

With customers plugged directly into these complex systems from the Web, almost everything has become mission-critical. To cope with this complexity, we have thrown a larger number of monitoring tools at all the different tiers: end-user experience, proxies, networks, servers, applications, message brokers, databases.

Yet companies are still losing millions of dollars due to unplanned outages and unacceptable performance. Plus, we are wasting days in war rooms trying to track down critical problems. If it works, we do not know why — and if it breaks, we do not know why.

Managing technology at the component level is no longer enough. Systems are too complex and dynamic. We are getting lost in the trees and not seeing the forest. When IT speaks, the business has no clue what it’s talking about.

(click on image to enlarge)

Transactions: The Business Context to Connect the Dots

The business and its customers don’t care about proxy servers, LDAP authentication, disk I/O, memory leaks, and message brokers; they care about successfully completing their activities and transactions. That’s what IT reliability is all about!

So, what if we were able to look at what goes on in IT and the data center from a business transaction perspective? What if we could put on the 4-D glasses of business transaction management and see the connections between the four dimensions of end-users, applications, infrastructure and business processes?

We would be able to know exactly where in the entire data center topology a transaction was getting stuck (what was slowing it down, what process it was waiting for …)

We could monitor in real-time the health of all transactions associated with a particular service or line of business (online banking, claim processing, new accounts, e-commerce …)

We could trace a runaway database query back to the user who originated it.

We could see the exact impact on all critical business functions of rerouting transactions to a disaster recovery site.

We would have a common context and language for business, IT, application development and operations to work together.

Case in Point

(click on image to enlarge)

Consider the example of a major investment bank whose online trading application was hanging for several minutes five to six times a day with huge negative repercussions for thousands of users worldwide.

Their IT experts convened in a war room but were unable to identify the cause of the problem and decided to try a business transaction management solution. They installed agents, and the BTM solution went to work performing a system-wide autodiscovery and aggregating data at the transaction level across all the tiers in the data center.

When the problem showed up again, they were able to see immediately that the bottleneck was in the external database. Zooming into a granular breakdown of the workload for that database, they found a specific SQL statement that was accounting for 90 percent of the transaction workload and was affecting the performance of the entire application.

However, identifying that SQL statement was not enough to fix the actual problem: The statement could have been sent out as a result of any number user actions, or a defect in the application code could have been causing tens of thousands of statements to be sent out to the database for no real purpose.

Armed with BTM data, the IT pros were able to find out which transactions were sending out these SQL statements and the user who was initiating them. It turned out they had a client who was an independent stock broker. He had two servers and had written his own code to “screen scrape” the bank’s application in order to collect market data to improve his trading of stock options. He was doing it in short bursts when there was a good opportunity for him in the market.

Right away, the IT team blocked the account and fixed the application code in order to prevent this from happening again. The problem was solved.

This example shows how critical a business transaction perspective is in order to provide the context necessary to understand what is going on in IT, plan proactively, and rapidly resolve even the most challenging of problems.

In a Nutshell

From managing incidents to managing changes, availability, security and compliance, BTM has been leveraged by large IT organizations to augment their ITSM activities.

If BTM makes so much sense, why has it been so difficult to implement? There are six fundamental challenges that a BTM solution must address:

• Monitor all components across all platforms that a transaction touches.
• Map all the hops of a transaction, beginning at the end-user, in a complete transaction topology.
• Rapidly implement the solution without invasive instrumentation.
• Do it without impacting the performance of monitored systems.
• Manage large amounts of data very fast and efficiently.
• Analyze data and present it in actionable, intuitive dashboards and reports.

To avoid pitfalls, a BTM solution should employ self-configuring agents that collect transaction data at the OS level. This is a noninvasive and platform-independent approach that can be deployed rapidly and has virtually no performance impact on production systems, allowing organizations to monitor all their transactions, all the time.

It should have the ability to “stitch together” the individual transaction data point into a map of the transaction topology and to monitor the metrics of each transaction for SLA (service-level agreement) compliance. Business analytics can then be applied to the transaction data to proactively manage services and accurately pinpoint problems.

Business transaction management will not replace traditional monitoring and performance tools that may be in place, but it will complement them and drive value by better pinpointing where they should be applied.