Mark Russinovich on Azure SQL Database Edge, Hyperscale, and beyond | Data Exposed

Mark Russinovich on Azure SQL Database Edge, Hyperscale, and beyond | Data Exposed

[MUSIC]>>Hi, I’m Anna Thomas, a Data and Applied Scientist
on the Azure SQL Team. In this episode, I’m joined by our
CTO of Azure, Mark Russinovich, who’s going to talk
to us a little bit about what we’ve been doing with SQL and what’s coming in the future. So thanks Mark for joining us.>>Thanks, Anna for having me.>>Yeah. So on the last episode, we talked more high level about Azure and data
and where we’re going, but I’m a SQL person. So I was wondering if we okay
if I ask some SQL questions?>>Sure.>>So the first one is
Azure SQL Database Edge. Now, I saw you present at Build back in May about
SQL Database Edge, and I was just wondering if you could share what your involvement has been and any insights you have
into what we’re doing now.>>Sure. Well, I think Microsoft’s worldview
as defined by Satya is Intelligent Cloud Intelligent
Edge and the realization of that worldview has actually built on the foundation which is Azure
including Azure Data Services. The goal is to be as consistent as possible from Cloud to Edge while delivering capabilities
that are specific to each requirements in each one of
those points on the spectrum. So if you take a look at data, it means delivering the same APIs
both in the Cloud and the Edge. So you can have an
application that can run in either place without major changes. Now, in the case of
Azure SQL Database Edge, there are some specific requirements that are related to
Edge that it fulfills. One of them being, you need
to take it and deploy it on an infrastructure that’s operated by somebody that’s
operating the Edge. It’s not going to be managed Cloud
service and so it’s going to be a container orchestrated
like Kubernetes which means that we need a containerize it. It also means that unlike the
Cloud which is where we run our Cloud services on
whatever hardware we want to and it’s not visible directly
to the customers in most cases, in the case of the Edge,
they might want to be using X86 or ARM for example. So we need to support both of those. The next one is on the Edge you’re going to be having
a bunch of data processing inside of one container with an integrated capability unlike the Cloud where you can decompose
things into different services. So for example, streaming, we want to have built right in. We also want the ability
to replicate data out to the Cloud as well for that
to be built right in. So all of these go into
that realization of that point on the spectrum of Intelligent Cloud Intelligent Edge
with Azure SQL Database Edge.>>Definitely, a really cool thing that we’re doing
and I’m excited to see what new applications it enables. You mentioned the Intelligent
Cloud Intelligent Edge and at Inspire I
heard Satya say that, Bill Gates was really impressed by Hyperscale and SQL
Hyperscale is one of the things we’ve been
doing in the Cloud and I know this is something that
came out of Microsoft research. It’s been something we’ve
worked on for several years. What’s your take on Hyperscale? What has been your
involvement and where do you think it’s going
and what does it open?>>Well, if Bill
Gates says it’s cool, it’s obviously pretty cool. The big innovation with Azure SQL Hyperscale is
the fact that it took this Monolithic database
architecture and decomposed it. Once you decompose it, you
can get to scales that you couldn’t get with the
Monolithic architecture. So taking the logs
door pulling that out, taking the compute servers
pulling that out allowing you to scale those out and to read
from the replicas directly, taking the page servers pulling those out and
allowing them also to scale, and then taking all the
persistent data and putting that on low cost backend Azure Storage
which is infinitely scalable. The combination of all that
means that you can get to 100 terabyte size databases with incredible amounts of
throughput because you define how much throughput you want on the read side with the read replicas. So really in a unique
way or new way to look at database architecture and
deliver this cool capabilities.>>Definitely. Speaking of new
ways and new capabilities, is there anything you’re
really excited about or new investments that we’re making
that you can share publicly?>>Well, going to that Intelligent
Cloud Intelligent Edge. I didn’t mention by the way that R is built right into Azure SQL Database Azure
ML right there on Edge. But that intelligence is one
part that we’re delivering. A cool part of intelligence
is not just giving customers the ability to
run ML right on the data, but having ML make a better service. So if you take a look
at the way that we’re doing things with Azure
SQL in the Cloud, we’re using Machine Learning to proactively determine if there’s going to be problems
with the database, to help customers
with their indexing, and just to optimize
the service overall. But one of the trends that
we’re going to be seeing is effectively customized
instances of Cloud services. Ones that are customized and
optimized for a customer’s workload or workloads in a way that
makes them perform better. So if you take a look at SQL, if you take a look and you’re
looking at the queries that workload is performing
rather than tell the database administrator
go put an index on this, organize, which only gets you so far. It is automatically determining which indices to put in
place and also laying out the indices in storage to make it efficient access for
the workload patterns. So I think that that’s going
to be a trend across not just SQL and databases, but just all Cloud services.>>Using the power
of Machine Learning.>>Yeah, ML for instance
specific optimization.>>Wow. That will be pretty cool. I’m looking forward to seeing it. Well, thanks so much
for joining us again, Mark. I really enjoyed it. I hope everyone else
enjoyed viewing this video. If you did, give it a like,
subscribe to our channel, and tune in to Data
Exposed next time. Thanks. [MUSIC]

Danny Hutson

1 thought on “Mark Russinovich on Azure SQL Database Edge, Hyperscale, and beyond | Data Exposed

Leave a Reply

Your email address will not be published. Required fields are marked *