I am trying to figure out if I can use the Data Access engine with L2 caching on a cloud application designed for Amazon that will entail autoscaling. NLB and MSMQ are not optimal solutions in this environment. At least from what I can figure, the easiest thing to do would be to disable the L2 cache, though I would hate to go without it. Retaining it means using MSMQ which is old tech and doesn't fit well into our architecture, at least that I can tell.
I know one of the first responses here is to use sticky sessions. Sticky sessions do not work well with corporate accounts as these guys have an army of proxy servers and a users request may be routed dynamically through any of them. I've been there already on a previous project that the solution (outside of Amazon at the time) was to use MSMQ to sync the servers.
Wouldn't it be grand if I could have my autoscale on Amazon and still leverage L2? Is this a dead-end?