The Register calls for aid, and Microsoft's Rohan Kumar will answer... our questions about SQL Edge and Azure Synapse

There was SQL Edge and Azure Synapse news at Microsoft's reimagined Build gathering this week, so The Register had a chat with corporate vice president Rohan Kumar about the company's database ambitions. SQL Edge Having lurked in limited preview for a while, Azure SQL Edge has been pushed out to a broader audience as the …

[bpfh]

Surely you mean...

Micros~1?

[jake]

Re: Surely you mean...

It's only Micros~1 when they aren't paying for the advert.

[Lorribot]

So we centralise in the cloud, then decentralise to the edge... then back to the DC. Then the edge. Then back to cloud.......

Does this run on Cisco edge switches?

Can I run my Data Warehouse on it. Will my half arse 3rd party developer (MS Gold Partner) of my LOB support this? No?

So its just another fancy piece of shit with a poncy name that only 1% of the real world will actually use.

[bombastic bob]

Micros~1 may be onto something here...

I'm in "wait and see" mode on this one.

DECADES ago I wrote a business analysis tool for Windows that (for whatever reason) didn't market well... but people who used it really liked it, saving them lots of time and letting them do ad-hoc analysis. The tool basically worked on some pre-crunched information locally on a PC, which allowed a number of ad-hoc reports and analyses on "the local data", including statistically derived stuff, etc.. It would have been difficult to manage that on the LIVE data.

So I "get" the concept. You pre-crunch some of the data in a way that you expect a LOT of clients/users/whatever to want, using an AI of some kind to figure out which should be pre-crunched or cached, and allow multiple "remote things" to report from the cache rather than the up-to-the-second live data.

This is also NOT unlike the reasons for DNS caching. So where appropriate, it's a good idea. AI would (in theory) allow you to get biggest bang:buck ratio.

So, "wait and see" mode. It might help, it might be cumbersome, and you're not gonna really know unless you try it out.

[another option, something else I've done before, is to cache any 'new' calculation results long enough to be re-used, and just age out cache entries to limit storage - this is simpler but less of a performance boost than a pro-active AI-based solution *might* be]

I would also NOT be surprised if many google searches have been pre-optimized in a similar manner.