One of the best solutions has got to be the HA and DRS portion of the vCenter where it's kind of an auto-load balancing and auto-recovering of your cluster if one of your hosts happens to die. Luckily it's pretty solid and you don't really have to deal with a lot of the HA stuff, but DRS is definitely very handy.
Room for Improvement
One of the things that I really wanted to see was the catalog because it came from vCloud Director, and they are adding that in 6.0, so they have that catalog, and they are extending it to where you can really replicate those catalogs out and share them. That was one of the features I would have really liked to see, and fortunately it's there.
One of the other features that we had been wanting to see was the vMotion between clouds, which of course that was announced today, that it's one of the things that's coming. I think that's going to be a game changer really.
Use of Solution
We have quite a lot of list of solutions that we use. Primarily we have a product called Server Virtualization, powered by VMware, and what it is, is it's a multi-tenant vCenter, but single-tenant hardware side of things. We have over 9,000 hosts in there and roughly 55,000 VMs worldwide, and six data centers globally. We also have other products called dedicated vCenter which is just that; customers get their dedicated vCenter. We also leverage vCloud Director; we have a dedicated vCloud Director product. We even have some other disaster recovery products that use SRM and vSphere replication.
In our environment -- it's pretty large -- we have roughly just less than 100 vCenters, and these things have to be up 24-7/365, and it's very rare that they actually go down. Out of all 9,000 hosts, I can count off maybe five that have crashed in that last year. That's a very stable VMware environment, in my opinion.
In a global environment with all six data centers, we have so many hosts and VMs, that we work very closely with VMware to help push their maximums, and really push the envelope of how many VMs we can fit in a vCenter, and how many hosts we can fit in a vCenter to really help drive those maximums even higher for VMware.
Customer Service and Technical Support
At Rackspace, we get mission critical support from VMware. If we have any issues, we call directly into VMware, and we usually get a response within four hours. Any major technical issues that we come across, they are very responsive; they work with us to help figure out what the problem is, and a lot of times we've found bugs in their software and helped them release patches to fix them.
Interoperability is one of the things I really like to look for: How well the VMware solution plays with other either hardware vendors, or other solutions, other service providers, other add-ons to make it even better. That's something we really look for. How well does it work in a service provider environment as well, because most service providers are different.
vCenter environment is super simple. Install ESXI; it's very straight forward. Then getting vCenter off the ground and going is extremely straightforward, especially if you use the vCenter server appliance: You can be up and running in a matter of a few hours.
We've set up a vROPS and we've found that it works really well in a service provider environment, that you can point it to multiple vCenters. We've worked with other vendors to do some deeper logging, deeper metrics gathering, and in doing so, we worked close with VMware and that other vendor and really built a full scale out worldwide global monitoring and alerting solution. That's one of the things we look for: Something that to fill a void that we don't have.
There is nothing else enterprise-ready, like they are. If you are considering similar solutions, make sure you take deep dives technically into how well they integrate with other vendors, or how well they integrate with your hardware. Like VSAN for instance, a lot of the storage vendors that are really going for it to be on the VSAN HCL, and unfortunately some of the RAID controllers are not on at that HCL, and a lot of times people don't know that. If you are looking at different solutions, make sure you check compatibility guides, not just for the whole VMware Stack, but including maybe subsets like VSAN, or other tools that you might be using.
We are looking at productizing that and making an offering for our customers, as well as using it internally. We've got it in several labs doing different things, and it's awesome. I really like it. It's resilient, in my opinion. A lot of people say if you only use the three-node minimum recommendation you might have data loss. I had a three-node cluster setup and my switch died, and when I replaced the switch and it came back online, everything was still running just fine, nothing had actually gone down; no data loss, nothing. It's actually really resilient. If you think about your data path, the data locality, it's a lot closer to the CPU, it's right there in that flash recache. It's a resilient storage solution that's cheaper than a dedicated SAN, or something of that sort.
Peer reviews are extremely important to me. I usually start Googling and looking to see, or on Twitter to find other vExperts, or other just subject matter experts that have talked about it: What benefits they've seen, or maybe pitfalls that they've seen. To me, that holds more water than a lot of the White Papers I've seen, because White Papers target maybe a specific use case, but I want to see more broadly: "How does it function? How does it integrate? How stable is it, of all things?" I really value the community involvement and opinion of others when I'm looking at solutions.
Disclosure: I am a real user, and this review is based on my own experience and opinions.
Nov 09 2015