Presented by Movere
When we started building Movere in 2010, we were 95 percent focused on IT discovery and inventory. We gathered performance data and usage metrics only when requested by customers. In 2013, our customers started asking us how to help them prepare for migrating to the cloud. Naively, we thought the inventory performance data and usage metrics already provided by Movere would ensure successful cloud migrations. It wasn’t until we had to perform our own cloud migration that we realized the error in our thinking.
Movere’s data warehouse is the lifeblood of our company. In the last 18 months alone, we’ve collected over 73 billion points of data from over 1,250 enterprise customers. To move our data center from on-premises to the cloud in 2014 required a rebuild of our entire environment not once, not twice, but four times!
Yes, I said four complete, painful, rebuilds of our entire environment.
When we first started our migration, we were excited about the cloud. Thrilled to be getting away from our already resource-constrained datacenters, we could now start implementing innovative technology that was exclusively based in the cloud to further optimize our platform. How ignorant we were to think this would be easy because we had good inventory data. After our first two failed attempts, we were resigned to building something, seeing how it performed, and if it didn’t deliver, we’d either re-size in the cloud or delete it and start again. It wasn’t until after the fourth rebuild, and multiple conversations with customers about their cloud migration experiences, that we realized these problems could only be solved through automation.
We made three missteps in our cloud migration journey:
- We failed to factor in the IOPS and throughput caps placed on each VM profile. We thought we had our compute- and memory-sizing dialed in relative to performance and cost, but we discovered we couldn’t get anywhere near the IOPS and throughput levels Movere was currently delivering on-premises in the new cloud environment.
- We failed to consider the age of the CPUs we had in our datacenter. We needed to identify the single-threaded compute capacity for each of the six processor models we were using in our datacenter, multiply that by their performance results, and then spin up the equivalent VM profiles we were focused on in the cloud.
- We failed to identify accurate memory usage. Hypervisors calculate ‘active’ memory using a sampling approach, calculating from the outside, using an approximation at the hypervisor level. This approach does not provide accurate data if previously allocated pages are re-used. So, even though memory usage is occurring, the host won’t be aware of all usage.
Our failed attempts at cloud migration became a defining moment for Movere as a company. We were able to learn through our own painful experience exactly what information our customers would need to accurately plan for their cloud migration. As a result, we needed to fundamentally evolve the Movere platform.
Movere shifted from an almost-exclusive focus on the discovery and inventory of IT environments to the collection of performance metrics down to the individual device level. This meant collecting thousands of performance data points from each device every few minutes and processing and presenting that data in almost real-time. We started collecting performance data at the processor level to help identify container/serverless computing opportunities. We sized for multiple clouds, so our customers had the data they needed to perform accurate comparisons. We focused on CPU, memory, IOPS and throughput, but we also factored in latency, cost, application dependencies, and hours of operation. We completely re-engineered Movere to solve a seemingly never-ending list of problems. Who knew we would be opening Pandora’s Box when we started the cloud migration journey?
What our experience taught us informed the development of a platform that could help our customers avoid the same painful learning curve we endured. By capturing performance data from the working set of each individual device and combining that with daily updates on cloud pricing, weekly cloud benchmark results, and leveraging a standard deviation to accommodate peak workloads, we provide our customers with all the performance data needed to emulate the most effective cloud migration strategy. Add application dependency mapping to ensure nothing breaks or is forgotten during the migration and, well, Movere provides just about all the information necessary to set off on your cloud journey.
Every day, command and control for IT infrastructure becomes more critical to the success of enterprise organizations. From planning cloud migration, to controlling cloud costs, to ongoing optimization and consolidation of environments, having continual, complete and accurate data in one location is critical to putting a lid on Pandora’s Box, and keeping it there.
For a free guide on what data to evaluate when planning a cloud migration journey, click here.
Sponsored posts are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. Content produced by our editorial team is never influenced by advertisers or sponsors in any way. For more information, contact email@example.com.