With the plethora of solutions and services available these days, where do businesses start? And how to best measure IT cost efficiency?
It starts from ground level. Before you even consider applications and other solutions, lay out your plans for a production-ready IT infrastructure. A production-ready IT infrastructure is one that has been performance-tested. It is one that provides a security-rich, available and recoverable environment. It must handle unplanned--not just scheduled--events. Those traits will form the basis for long-term business expansion and growth.
An IT infrastructure is built using an integrated stack of different layers. They include the following:
- Business applications Application server/database platform
- OS platform
- Logical Software layer for storage (file systems)
- Server hardware
- Storage drivers, SAN fabric and storage.
In most cases, these issues are a result of skill- and product-mismatch. Typically, organizations adopt a best-of-breed approach that invariably involves different vendors in the implementation of their IT infrastructure. However, hardware and software vendors, with their different knowledge and skill sets, are typically adept at only focusing on their areas of expertise. Although much standardization effort exists in the industry today, competition drives each vendor to release products at different times.
This can lead to potential issues of compatibility--with integration, performance and security. And IT managers should be mindful of these issues which may arise during the building and deploying of their IT infrastructure.
An eight-point deployment plan
So how can organizations achieve a smooth implementation of a production-ready infrastructure? To do that while minimizing loss of business time and revenue, consider the following approach:
Phase 1: Design
Design the IT infrastructure solution for your business application needs using a proven reference architecture--one which is tested for performance, availability and security. Make sure the testing is a joint effort by the different vendors involved.
Phase 2: Build
Implement the above reference architecture with the requisite customization to meet your business-driven requirements. Then follow that with the implementation of best practices at every layer of the integration stack. Again, consider the recommendations of each of the different vendors which are involved. Focus on interoperability, performance, availability and security-related recommendations.
Phase 3: Test
When it comes to testing your architecture, aim to have a different team perform the quality assurance. During the testing phase, conduct a unit-level test of each integration stack of the IT infrastructure. While quality assurance should be part of the requirement analysis team, it should be kept separate from the implementation team. This ensures that standard test procedures and process are conducted for the system under test.
Phase 4: Integrate
Conduct a system integration test (SIT) to meet the requirement specifications of the system. Be very specific with your requirement specifications. Be exhaustive--cover every aspect of the system design if possible.
Phase 5: Analyze
Perform a load test according to the design capacity of the systems on a standard benchmark application. This is the time to identify system configuration issues or hardware bottlenecks. Once problems are spotted, aim to eliminate them as early as possible.
Phase 6: Deploy
Start deploying your business applications on top of the IT infrastructure that has been built and tested so far.
Phase 7: Audit
Perform a security audit on the environment using best security practices and taking into consideration corporate policies on security.
Phase 8: Completion
The final phase of the build-up is meeting the business users' service-level agreement (SLA) requirements. An SLA is typically worked out in terms of a number of transactions for a defined number of users within 99% of your specified timeframe.
At this stage, you should perform a load test with application performance analysis tools to see whether the defined SLA requirements are met. If they are, great! But if not, don't put off remedial actions. Endeavor to drill down your analysis on bottlenecks that can exist in network, servers or storage. If necessary, drill down to the code level of the applications.
Unfortunately, or fortunately, your work does not end with the meeting of SLA requirements. As a practice, conduct performance analysis so that you can document the areas that take more time in the entire build-up. This serves as a useful reference for future implementation and maintenance. It also helps your organization optimize its resources, thus saving on the IT infrastructure cost.
In the same direction
With a core plan in place, organizations with large IT teams of various expertise and different focuses can have a better picture of how they can better work with one another. This helps to align them with the overall needs of the organization, thus enhancing operational and business efficiencies.
Gowthaman M is the technical director of Frontline Technologies, a Singapore-based regional IT service provider, where he leads a team in providing business solutions, professional services and engineering. A 15-year IT services and operations veteran, Gowthaman's current interest is in the area of tech investment optimization