Couple of questions prior to signup

March 11, 2013 3.9k views
What kind of hardware redundancy is in place? Are all the ssd's local storage or is it on a nas\san storage? Are these rackmount servers or blades? What kind of density is per host? What kind of guarantee do I get that the host is not going to be overloaded/oversold?
5 Answers
The SSD's are local storage and they use KVM which generally prevents RAM from being oversold. CPU is assigned per Droplet and from my experience, you're allow to use the full CPU Core allocation.

We've frequently fully utilized 2-8 CPU Cores (depending on the Droplet) without any issues and speeds have been great.

I can't really comment on density, though they have done a great job making sure that other clients don't affect our service. We've had one issue with a "noisy neighbor" though it was handled extremely fast and since then we've not had any issues with other clients affecting our service or our clients.
Local storage means NO HA. what happened in case of failure of one host.
What is RAID Level 1,5,6 or 10? Data is very important, not sure in case of failure it is replicated on other storage or not..
All servers run hardware RAID with different RAID levels but in each case it requires multiple drive failures at the same time for any data loss to occur. In most cases when there is an issue usually a system becomes degraded and starts to exhibit anomalous behavior, at which point we determine that it is no longer healthy to keep it in production and open a maintenance window with customers and migrate their virtual machines to a new hypervisor.

In an emergency case we may move the virtual machine without much prior notice if we believe it is imperative to clear out the hypervisor sooner.

In regards to data storage we highly recommend to use several different levels for your backups. That includes enabling the internal backups that we provide, taking the occasional snapshot, and also implementing some sort of off-site backup yourself where you focus on moving the content and data that is the most essential to you.
How you guys migrate your VM's from other hypervisors if your provided VM's run on a local storage... one point we can understand the VM's degraded their performance but how they migrate?

Another question please clarify this "The SSD's are local storage and they use KVM which generally prevents RAM from being oversold" means r u guys using SSD's for OS or used as for replacement of RAM??
DO doesn't use SSD for RAM, only disk storage. The SSD's are local to the hypervisor, the droplets are allocated inside the hypervisor so to speak.

I believe what the support is saying is if a RAID array in the hypervisor degrades too quickly to recover normally (by replacing a single drive for example) DO will move the droplets off the hypervisor to another one.

At the end of the day, I always take responsibility for my own data security. DO's setup is very good, much more reliable than a dedicated server, but any comprehensive data integrity strategy must include steps on your side as well as they suggest.
Have another answer? Share your knowledge.