I have to admit, I approached the whole question yesterday quite differently that everyone else. I was worried more about new standards, development trajectories, compatibility and so-on. When I wrote about the vendor-specific database flavors… I missed what could certainly be the larger issue.
John Shadows wrote:
100% cloud also means you have to play by the cloud rules all the time.and ahazelwood wrote:
The biggest drawback is the 100% tie to the cloud.
Which got me thinking. One of the bigger gotchas in the cloud is vendor lock-in IMHO. Vendor lock-in is what makes it really difficult to move between cloud providers. Sure, DNS is DNS. But the WAY one is set up and that you’re used to and the little things that are offered with it to make life easier – those can be challenging to recreate and re-learn. The same is true for the database implementations. It may still be SQL Server, but the implementation and the hoops you move through to deploy mean a good deal of lessons learned would be set aside and restarted if you move to a new platform.
Throw in there a new, vendor-specific database and the things could get even muddier.
Part of me is all “throw caution to the wind – damn the torpedoes, full speed ahead!” when you pick a vendor, just go for it, use their tools, their environment, their systems. But that lock-in is a very real thing. We do some very cool stuff on Amazon that makes our lives much easier. Automation pieces, transcoding, some funky database stuff. Those are all things that tighten the bolts that hold us in at Amazon. We’ve been considering a move to Azure and find that it’s very difficult to do wholesale, but possible in pieces. That fact gives us pause in the process.
That’s not a fault of Azure. It’s a fact of life. The same is true if we wanted to move to on-premises servers, or a mix of on- and off-premise solutions.
There are great things to be had in the automation and management and control you get with the Azure dashboard and very deep functionality in the platform on the whole, with the tools on AWS, etc. But vendor lock-in has to be a thing, it has to be a consideration with every single choice you make to deploy a new function or capability.
It doesn’t matter if it’s an AI-powered chat solution (they all have them, they’re all different and have different nuances) or a standards database or a proprietary database.
I have to admit we’ve shy’d away from some solution ideas because they were just TOO new and evolving and vendor-specific. We did this in hopes that some standards would evolve and that it would make more sense and be a bit of a more mature offering if more players were involved and pushing and shoving to get things really solid.
We’ll have to wait to see how that goes.