Travell Perkins

From the communication that we have received from MT it looks as if the storage solution is the Achilles heel of the system. Could we get some documentation on what exactly that storage solution looks like with a topology map? From the description so far it would appear that a NAS based solution was implemented, where the NAS server was some huge beast attach to RAIDS. The current fixes probably replicated this beast and assigned pools of customers to each one.

How about clusters of NAS servers backed by a SANs? Each NAS cluster would have a SAN. Wouldn’t this hybrid solution be easy to scale in multiple dimensions? I’m just guessing here because I don’t know the details, but many of us need more transparency. I got up today feeling good about launching my new service on the grid. Then I got slapped with a latency spike. Also, does storage latency effect MySQL Containers?

I’m sticking with MT because I know the general design scales so that eventually a request will go through if it doesn’t time out (rare occasion for me at least). I also feel that these issues can be resolved, but you can’t do it with open source. You need to lay down some money for your storage systems. The architecture is the hardest part, but once you have it down it’s smooth sailing. Looks like the architecture still isn’t perfect.