PASS 2015 Session Report – KeyNote and SQL Server on Azure Virtual Machines: Features, Best Practices & Roadmap

So PASS officially kicked off this morning leading into the next 3 days of back to back sessions.

You could certainly tell that the keynote was on… I mean the dining room was pumping…!

WhereTheBloodyHellAreYou

 

 

Oh that’s right, everyone is at the keynote!

KeyNoteDay1

 

 

So the Keynote session was hosted by Joseph Sirosh Group Vice President, Data Group.

The big tell for the key note was undoubtedly the SQL Server 2016 CTP3 and just whats packed to the rafters within the software.  If you want to learn more about that then I recommended step across to this link here http://blogs.technet.com/b/dataplatforminsider/archive/2015/10/28/sql-server-2016-everything-built-in.aspx

Key Takeaways from the Keynote;

  • SQL 2016 is really a major release that really solidifies the Microsoft view of a solid foot in both the On Prem and In Cloud data platform camps.
  • “The future is both earth and sky!”
  • The release offers much On Prem capability like Polybase (to APS), R integration (advanced analytics), Always Encrypted, SSAS/SSRS improvements
  • The release also provides the ability to seamlessly integrate from On Prem to Azure Cloud – and/or back like Polybase (to HDInsight), Stretch Database – and SQL already has capability to use Azure VM’s for SQL AAG solutions and Azure backups.
  • An interesting takeaway – the human size of human genome is approx 1.5 Gigabytes, or about 2 CDs worth of storage space.  How small do you feel now?

 

I then attended 4 sessions, but today there is really only time to blog about this one, mostly for me it was the most impressive in regards to capability and just how far its come!

The session was SQL Server in Azure Virtual Machines – Features and Best Practices and was presented by Luis Vargas is a Senior Program Manager Lead in the SQL Server team.

Key Takeaways from SQL in Azure VM’s

  • This was a great session – it has been a little while since I looked at Azure VM’s in some detail – and I am blown away by just how far the technologies have progressed.
  • A straw-poll of the room indicated that 90% of attendees run SQL in VM’s today
  • The orchestration and deployment is seamless – it was very easy to spin up a new VM of a set size and have SQL Server pre-installed onto it.  There was even an amazing demo to deploy an entire set of Azure VM’s participating in a 2x node SQL 2014 AAG solution, including a set of AD Domain Controllers (in case you didn’t have your own AD already!)
  • The owner of the VM is responsible to patch the VM Guest OS and SQL itself
  • When you deploy a VM with SQL the cost includes the VM run-time (to the minute usage), Network out (not in), Storage Used and SQL licencing.  You can Bring Your Own Licence (BYOL) if you have a Microsoft EA – that’s for SQL!
  • SQL 2016 CTP2 is a standard image available for deployment – and SQL Express will soon be an image that can be deployed – obviously scaled down however its free of SQL licensing costs.  All images come with the latest CU within a few weeks since the CU was released.
  • Every VM disk is a BLOB in Azure Storage.  There are 3 local sync copies of that storage and 3 more async copies if remote geo is selected (and paid for!) too
  • There are two types of storage; Standard (ie HDD) or Premium (ie SSD).  Naturally the SSD absolutely smashes the IO @ 200MB/sec per disk with a 4ms latency.  Who does spinning disk now-a-days? 🙂
  • There are a number of preconfigured different sizes from 1 core and 750MB of RAM running SQL Web Edition through to 32 cores and 448GB of RAM running SQL Enterprise.  NOTE that not all configurations are available in all Regions.  A quick calculation at Year 1 shows me that its easily cheaper running the Azure VM with the SQL licensing bundled in as opposed to actually owning and deploying this within your DC, though you would need to extrapolate this out over say a standard 3 year life-cycle to compare apples and apples – but I still suspect cheaper considering no hardware maintenance/upgrades and instantaneous scale up/down options should your workloads change.  Also if you shut down the VM then you dont pay for the compute!
  • Azure Storage (ie the VM disk BLOBS) guarantees no data loss
  • Can configure Azure Availability Sets (NOTE different from Availability Groups) which is kind of like anti-affinity in that it ensures that 2x VM nodes within the same HA/DR solution such as SQL AAG will reside on different racks in the underlying infrastructure.  This obviously protects the solution should a rack/host fail.
  • Azure VM’s provide the ability to auto-patch for both Windows and SQL – nice!  You can turn this off if needed too
  • Azure VM’s provide the ability to backup the SQL databases to VM Disk or to Azure Storage itself (which means being accessible from SSMS, etc).  There is also an automated backup capability which backs up based on actual database usage (given transaction log record generation) – would like to see how that works for real
  • The entire solution can be monitored by Azure Operational Insights Portal (think System Centre Operations Manager, but in the Azure Cloud).  It checks over 123 SQL best practices and makes recommendations for improvements.  Sweet!

Overall simply great Day #1!

I am running my session tomorrow on Using Azure Machine Learning to Predict Seattle House Prices – so I hope that it goes well!


Disclaimer: all content on Mr. Fox SQL blog is subject to the disclaimer found here

Advertisements

One thought on “PASS 2015 Session Report – KeyNote and SQL Server on Azure Virtual Machines: Features, Best Practices & Roadmap

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s