Evaluating product features, comparing prices, and doing proofing of concepts are important steps in the process of adopting almost any new product. But once one completes those steps, the time arrives to start to roll the product out and implement it. In this second installment of my interview series with System Architect, Fidel Michieli, he shares how his company gained a comfort level with Cohesity for backup and disaster recovery (DR) and how broadly it decided to deploy the product in the primary and secondary data centers.
Jerome: How did you come to gain a comfort level for introducing Cohesity into your production environment?
Fidel: We first did a proof of concept (POC). We liked what we saw about Cohesity but we had a set of target criteria based on the tests we had previously run using our existing backup software and the virtual machine backup software. As such, we had a matrix of what numbers were good and what numbers were bad. Cohesity’s numbers just blew them out of the water.
Jerome: How much faster was Cohesity than the other solutions you had tested?
Fidel: Probably 250 percent or more. Cohesity does a metadata snapshot where it essentially uses VMware’s technology, but the way that it ingests the data and the amount of compute that it has available to do the backups creates the difference, if that makes sense. We really liked the performance for both backups and restores.
We had two requirements. On the Exchange side we needed to do granular message restores. Cohesity was able to help us achieve that objective by using an external tool that it licensed and which works. Our second objective was to get out of the tape business. We wanted to go to cloud. Unfortunately for us we are constrained to a single vendor. So we needed to work with that vendor.
Jerome: You mean single cloud vendor?
Fidel: Well it’s a tape vendor, Iron Mountain. We are constrained to them by contract. If we were going to shift to the cloud, it had to be to Iron Mountain’s cloud. But Cohesity, during the POC level, got the data to Iron Mountain.
Jerome: How many VMs?
Fidel: We probably have around 1,400 in our main data center and about 120 hosts. We have a two-site disaster recovery (DR) strategy with a primary and a backup. Obviously it was important to have replication for DR. That was part of the plan before the 3-2-1 rule of backup. We wanted to cover that.
Jerome: So you have Cohesity at both your production and DR sites replicating between them?
Jerome: How many Cohesity nodes at each site?
Fidel: We have 8 and 8 at both sites. After the POC we started to recognize a lot of the efficiencies from management perspective. We knew that object storage was the way we wanted to go, the obvious reason being the metadata.
What the metadata means to us is that we can have a lot of efficiencies sit on top of your data. When you are analyzing or creating objects on your metadata, you can more efficiently manage your data. You can create objects that do compression, deduplication, objects that do analysis, and objects that hold policies. It’s more of a software defined data, if you will. Obviously with that metadata and the object storage behind it, our maintenance windows and backups windows started getting lower and lower.
In part 1 of this interview series, Fidel shares the challenges that his company faced with its existing backup configuration as well as the struggles it encountered in identifying a backup solution that scaled to meet a dynamically changing and growing environment.
In part 3 of this interview series, Fidel shares some of the challenges he encountered while rolling out Cohesity and the steps that Cohesity took to address them.
In the fourth and final installment of this interview series, Fidel describes how he leverages Cohesity’s backup appliance for both VM protection and as a deduplicating backup target for his NetBackup backup software.