New task - plan for scale_out of server_architecture
I Need a bit of help, please: My Supervisor gave me a new Task - in the Long run, I'm supposed to be more into "ops" than "development", I don't yet know what to think of it, but I take it on for now.
My next Task is one I don't have any experience with because with my last employer I had no Chance to gather any: Server_architecture (we had a "Workshop" with somebody from Qlik on that last week or so, but that was of course quite general) - we as a Team are currently growing and taking on new Tasks and becoming more important, plus our existing QlikView_Server is at its Limits, so we are planning to use the Budget we have to scale out and distribute the QV_services on a broader Basis.
Specifically, after gathering some cost factors, we have decided to have a three-machine Setup:
- One for the QVS (quite big, that would be the one we are currently using)
- One for QDS, Nprinting and QEMC (a smaller one)
- One for an IID_Server (security) (a very small one as this is only - well, small)
=> Now my Task is to find out what should be the next steps to take to realize this. That is where the "I have no experience" Comes in 😉
Can anybody offer some help here? In the short run, I would be grateful for that (I guess that at some Point we'll Need a Qlik_consultant, no?). In the longer run, of course, I will happily read up on the issue if you could Point me at something useful (I have the ebook "QlikView Server and Publisher", but no more since this was not relevant for me yet).
I am already well into reading that "Server and Publisher" ebook and I'm beginning by writing out all important keywords, all things that have to be considered in some way. I can cleanse and trim my notes later. There is a lot to learn in that respect.
- We won't Need anything to do with Clustering as there will be only one QVS;
=> So we won't Need a load-balancer either, will we?
- I have taken note of a number of things the Server should have or not have to be well suited for QlikView - unfortunately we cannot take influence on a number of things (like the "overprovisioning" of VMs or the NUMA/ non-NUMA scheme)