Pythian Blog: Technical Track

New features and service offerings from Microsoft Azure - September 2018

by Warner Chaves on Oct 11, 2018 12:00:00 AM

I recently joined Chris Presley for his podcast, Cloudscape, to talk about what’s happening in the world of cloud-related matters. I shared the most recent events surrounding Microsoft Azure. Topics of discussion included: Business-critical service tier in Azure SQL Database Managed Instance now in preview Azure SQL Database features going GA:
  • Vcore pricing model
  • Auto-failover groups
  • Configurable long-term backup retention
  • Reserved capacity pricing
  Business-critical service tier in Azure SQL Database Managed Instance now in preview For people who are following the Azure Managed Instance story, the General Purpose service tier of Azure SQL Database Managed Instances came out in preview a few months ago. As soon as the preview started, it absolutely depleted all their preview resources. They had so much interest from people who want to migrate to Managed Instance, because it’s easier to migrate to if you're coming from SQL Server compared to the existing Azure SQL DB servers. Now they have introduced “Business Critical” which is going to be the premium tier for Azure SQL DB Managed Instance. The first difference is that General Purpose uses remote premium storage compared to local storage for Business Critical. General Purpose is still SSD, but it’s remote SSD. Business Critical uses local NVME SSD storage. It’s a lot faster, it has lower latency and it has higher IOPS. Because it uses local NVME SSD storage, they’re also enabling in-memory OLTP only in the Business Critical service tier. If you’re on the General Purpose tier, you can’t use in-memory OLTP because the Microsoft team did some testing and decided that it really didn’t do the feature justice if you were using remote premium storage. Another difference is that General Purpose uses Azure service fabric to act similar to a failover cluster. If you lose your instance, then the Azure service fabric finds somewhere else to initialize your instance, starts it up again and then you’re back in business. Even though there’s an automatic failover, an HA provided there, it is not as fast as the one that you’re going to get with Business Critical. Business Critical is building it on top of availability groups and is keeping warm standbys at all times so failover times, in case of some sort of hardware issue or a VM issue on the Microsoft side, is going to be a lot faster. It probably won’t even be caught by most applications because it could be dealt directly by the driver. The last difference is that Business Critical is going to have read replicas so you can scale your read-only workload. For a deeper dive into this, read my blog post from August entitled Introduction to Azure Managed Instances. Azure SQL features going GA The features I’m going to cover next are in the single database model. Lots of activity happening here as well! Vcore model The biggest change is that the virtual core model is now available on Azure SQL DB. If you’re familiar with the service, you know that they used to use an abstract performance unit called database transaction unit. The database transaction unit was supposed to be a blend of CPU and memory and IO capabilities and it was just a flat integer number. It started at, I think five, and then it would scale up to a few thousands at the highest premium tier. But there was never any public disclosure regarding how many cores you get when you have these many DTUs and so on. The idea was that people shouldn’t care about how we’re configured and they should just care about the raw numbers that they get in terms of throughput. I don’t know if the problem was that this concept never really caught on or if they figured out that it was really difficult to give people good migration deals or discounts from what they had already purchased on premises. But they have introduced a new model where you can select the number of virtual cores and you will be charged by that amount. You get a certain amount of RAM based on the number of virtual cores. This is going to make sizing a lot easier for some people because they’ll just say, “If I’m using 12 cores on-prem, I want 12 cores in Azure SQL DB.” The other thing it does is it allows you to do license mobility. What’s going to happen is if you have a certain amount of SQL server licenses that you already bought on-prem and you don’t want to migrate them straight to managed instance, then they’re going to tell you, “Okay, you can move this many cores into the individual model,” and probably provide you with some sort of multiplier and give you a way to keep that license software assurance value on the way to the cloud. You’ll get a discount because you already had an existent SQL purchase and you’ve been paying for software assurance. Microsoft is just sweetening the pot to try to make people move to the cloud and keep the investment that they’ve already made on-premises. Auto failover groups The next one is Auto Failover Group. This is very similar to availability groups if you are familiar with that feature of SQL Server. You can group different databases and make them failover together from one region to the other. Azure SQL DB also provides one global endpoint so you don’t have to switch connection strings. Configurable long-term backup retention Configurable long-term backup retention is now GA, as well. Previously, the longest backup retention was a month and if you wanted to have something longer, you had to roll it out on your own. Now we have full long-term backup retention as part of the service itself. Of course, if you decide that you want to keep seven years’ worth of backups - potentially terabytes of data – well, then you are going to get charged for those. But if you do want to do that, you don’t have to roll your own solution. You can now just do it as part of the service. Reserved capacity pricing They have been doing Reserved Capacity Pricing with virtual machines for a very long time. Now you can do it with databases as well and get better pricing. What I suspect will happen is a lot of people are going to use the license mobility that they have with SQL on-prem, plus the fact that they know that they are not migrating anytime, and just combine both of those to get the biggest savings that they can. Smart move to reward loyal customers.   This was Part 1 of the Microsoft Azure topics we discussed during the podcast. Chris also welcomed Greg Baker (Amazon Web Services) who discussed topics related to his expertise. Listen to the full conversation and be sure to subscribe to the podcast to be notified when a new episode has been released.
Topics: Azure Technical Track

No Comments Yet

Let us know what you think

See all

See all

Subscribe by email
Follow us
Love Your Data
Pythian Logo

© 2024 All rights reserved.