I had the opportunity to test another backup and DR solution recently. This time, its Rubrik's competitor Cohesity. If you miss the Rubrik overview, you can search for 'Rubrik' in the search field located on the right. From the looks of the hardware (model C2300), it looks similar as the Rubrik device we've evaluated from a standpoint of each having 4 nodes blades with (12) 3.5" disks condensed into a 2U device but with a different faceplate. After having looked at Rubrik, I was excited in finding out what Cohesity has to offer. My initial thoughts, without actually diving into it, was that these two units does the same thing in the same way. After playing with it for a couple of days, I was correct. There are a lot of similarities between the two and some minor differences that does not have anything to do with what it advertise for. They both do backups based on taking a vSphere level VM snapshot. For replication, Cohesity copies the snapshotted VM over to a second appliance. And for DR recovery, it would just bring up a copy of the VM that got replicated to the second device. As with Rubrik, one of their main selling point is that your can bring up a DR instance within seconds by the virtue of mounting the Cohesity device to our ESXi host as a datastore and powering on the VM from there without actually having to do any restores. One thing that Cohesity is different than Rubrik is that Cohesity is job and policy based whereas Rubrik is SLA based. I much prefer Cohesity's job and policy bases over the SLA base because I can see all the VM's failed in one job in a single pane of glass view. Can't do this with Rubrik unless your run a report. Cohesity does offer you to create seperate storage domain so you can see how much a job uses because the reporting is storage domain based. By doing so, you get less deduplication as it does not dedupe across other storage domains. Some people wouldn't like it to have multiple storage domains but I find it the only way to see how much a space a particular job with Cohesity. Like Rubrik, Cohesity is very scalable. If you need more storage, you would purchase another node and add it to the Cohesity cluster. Done. To go further into the backup realm, Cohesity has the ability to archive backups to a third party cloud, able to backup up a network folder, and SQL database level backup using an agent. We found that the cost for purchasing Cohesity was suprisingly low. For a C3400, which provides a usable capacity of ~40TB (don't exactly remember the actual number), the cost was less than $90k for 2 units, with 3 year maintenance contract.
 |
| C2000 series |
 |
| 4 node cluster |
Pros:
- Can do VM backups, replication, DR and archive.
- Surprisingly low cost.
- Storage domains.
- Easy to scale.
- Job and policy based.
- Very granular retention policy.
- Can archive to third party cloud.
- Can back up network shares and SQL databases.
- Active directory integration.
- Good detailed audit logging.
- Easy upgrade process.
- Can mount itself as an ESXi datastore and run VM's off it making RTO time very low.
- REST API
Cons:
- vSphere level snapshot for backups. Can be problematic for VM's with large disks depending on what application it is running.
- Hardware based. Will need to recycle hardware when it gets old.
- RBAC permissions didn't work too well.
