Price New ATLA Server Config

[orangewolf]

Get cost data for converting ATLA from k3 to eks

[orangewolf]

I don't have access to see the whole CostExplorer, so I probably do not have everything on the stack in the price estimate so I'm just focused on the compute here. This allows us to see the relative costs pretty well. By moving to EKS we can either get significantly more compute power (about 50% more) for an extra $60 a month or we can keep the compute the same but save money each month. We can likely hit in the middle if we switch up the instance types, but that does make the set up a little more complex. All of this is without any pre-buy, which I'd advice doing once we make this shift. With a pre-buy you can likely get the extra compute and pay the same amount. I put all 3 estimates in one calculator because I think it is easier to look at this way. It is split up in to current set up / plan A / plan B.

[orangewolf]

AtlaCostEstimate.pdf

[orangewolf]

@crisr15 can you take this across the finish line from here? I'm happy to do a call if Christy or Jim has questions.

[crisr15]

@orangewolf What does the migration look like? How long would moving everything to any of the options take?

[crisr15]

Jim's response:
Jim is back today and he is saying to go ahead with the DL changes but temporarily, and that later he and rob can have a cost review conversation to determine what is best. Does this make sense on your end? I need the importer working properly to fix a collection that is being presented at conference on Thursday morning - so I think that is why he is saying temporarily.
follow up that maybe clarifies a bit more (from Jim) "I figure they could always up the instance type (to get more compute) without making any changes to the systems. But, if they are confident with transitioning to EKS, without the need for significant retooling, I am certainly not opposed to that. I also agree; we will want to explore prebuying resources once they know what is needed."

[orangewolf]

go with 3 m5.xlarges on eks