Mike Moore, Chief Information Officer –
When you’re making a technology solution for customers, it pays to use that solution yourself. You can see its strengths and weaknesses, and then work to build up the former while fixing the latter. In the industry, this is called “eating your own dog food”, a phrase that implies many tech companies are gagging all the way through this exercise. However, when the technology is strong and it solves a throbbing pain point, that “dog food” becomes a treat delicious enough for both the customer who buys it and the company that makes it.
At Virtual Instruments, we once faced many of the same infrastructure performance management (IPM) challenges that affect our customers, albeit on a much smaller scale. At times, we have grappled with:
When our customers experienced these same problems, we recommend they start with performance remediation, rather than get stuck in a pattern of reactive troubleshooting. That’s what Virtual Instruments did, too.
We rolled out our first SAN storage in late 2013 and chose Fibre Channel as the protocol to support the performance and throughput requirements of our rapidly growing organization. The primary users of that storage are corporate IT, support, and research and development. During the implementation, we migrated many bare metal servers to ESX to improve manageability.
To manage our infrastructure, we instrumented with VirtualWisdom. The first phase of our project, like many of our customers’ projects, was implementation and discovery. We learned quickly that a slow performance issue stemmed from improper configuration of the tiered caching on the EMC VNX. We also saw CV errors on deployment, improper VNX link setup, numerous multi-pathing issues, and buffer-to-buffer credit congestion.
All of this visibility led to informed performance remediation rather than troubleshooting in the dark.
Visibility = savings
The Virtual Instruments professional services team ran a copy of our VirtualWisdom server in the services cloud and saw that we had underutilized CPU and I/O, and we were maxing out on memory. Based on that information, we made some shifts that resulted in a 20 percent increase in capacity, with no increase in cost. We also learned a lot about our individual virtual machines, which helped us increase utilization of existing servers and therefore put off the purchase of new ones.
All of this eating of our own dog food gave us product insights that now benefit our customers. We’ll have another helping, please.
Request a VirtualWisdom demo to learn more about what infrastructure performance management can do for your business.