In many ways, the world of IT is still immature. We respond faster to hype than the average fashion addict. We’re often driven more by emotion than facts; the bully yelling the loudest gets whatever he or she wants (new servers, new architecture, etc.). Have you been to a high school reunion lately? Have you noticed what became of the class bully vs. the quiet, smart kid with the specs in the front row? Often surprising, isn’t it? In retrospect, given the choice, who would you respect more and seek out for advice or friendship?
At the reunion we must all admit feeling at least a touch of schadenfreude (pleasure derived from someone else’s misfortune) when we find out the bully ended up as a used car salesman. Everything is so much easier in hindsight.
But imagine you could re-examine some of the common IT decisions made in the past and choose again. For example, six or seven years ago, it seemed like a good decision to copy production data (IBM DB2/CA IDMS/CA Datacom, etc.) to distributed systems for data warehousing and other purposes. Mainframe MIPS were considered expensive vs. distributed servers (UNIX/Windows), so the decision seemed obvious. Looking at it today, we know that the average company will copy their largest production databases three to five times each 24 hours, to dozens of different servers. That’s a lot of data duplication, network usage and processing power, and, of course, it all involves many servers (which, of course, must be managed by a lot of administrative people).
Now, knowing this scenario is the alternative to simply copying the data to a separate mainframe partition does impact the choice, right? The ability to more easily compare the TCO, along with projected uptime, reliability, and disaster recovery capabilities, would cause most to choose differently, if they could.
Consider a recent production example: Until two years ago, a leading Minnesota healthcare provider ran SAP on Intel (with IBM DB2 on the mainframe). They were converting to a new release of SAP that required a number of additional servers. The proposed hardware costs prompted them to perform a five-year TCO analysis for server-based and mainframe-based solutions. They chose their new SAP platform based only on the numbers—what would be more cost-effective. No politics, just the facts. For them, the Linux on System z platform had by far the best numbers over a five-year period. The hardware savings were considerable because z/VM allows them much better sharing of resources among Linux servers, even vs. Intel-based VMware. They also aren’t restricted to sharing resources just among SAP servers, which they would be under Intel-based VMware. They even factored in the IBM System z model in which capacity doubles or more with each upgrade at no additional cost. Beyond the cost savings, they know their Linux on System z platform has lowered their risks and improved the service they provide to internal and external customers. They’re now two-plus years without a platform-related outage, have decided to use Linux on System z for their IBM WebSphere environment, and are evaluating each new and existing application to see where it runs best. They chose to re-evaluate an old choice and make an educated decision about how to move forward now.
Why not use the current economic climate to propose some changes for the better in your data center? This is a time when every potential proposal is taken seriously by management. Perhaps now is the ideal time to admit we’ve made some wrong decisions (and I don’t mean that tattoo!)? Take advantage of the current climate where positive change is OK, and start acting like adults.