T-SQL Tuesday #82 – To the cloud… And beyond!!!

T-SQL Tuesday is a recurring blog party, that is started by Adam Machanic (Blog | @AdamMachanic). Each month a blog will host the party, and everyone that want’s to can write a blog about a specific subject.

This month the subject is “To the cloud… And beyond!!!”. If you want to read the opening post, please click the image below to go to the invitation.



 
As I mentioned in the invitation, Azure allows us to easily host and scale our platform. But it also comes with a whole new set of challenges. Some of the observations I made in the last few months:

 
Agent is missing in action…
One of the biggest challenges for me when I started migrating stuff to Azure was the missing SQL Server Agent. I never thought about how often I actually used it until I missed it! There’s no more “let me quickly schedule this script for tomorrow morning” anymore.

And although there are multiple solutions to this problem available in Azure (Scheduler, Automation) I picked the easy way out. Because we’re running 2 platforms in parallel (an on-premise private cloud and Azure), I decided to create a VM in the private-cloud platform to run operational tasks like scheduled maintenance for example. This is not a solution I would like to keep for the upcoming years, but it allows us to easily run scheduled tasks, without losing time on figuring out how Azure Automation or Scheduler works.

For more information on this, please read this post.

 
Contained datastore
A thing that can make migrations to the cloud a bit more difficult, is that Azure SQL databases are basically a contained datastore (you would call it a “contained database” when you run it on-premise). This means that you (by default) can’t connect from one database to the other. This could mean that you need to rewrite your applications or stored procedures, or maybe even redesign your entire database/application/domain model.

This also means that running a stored procedure from the Ola Hallengren’s maintenance solution can only be done on the specific database, and not from the master database like the on-premise version does. These small challenges can be overcome, but it does mean code-duplication in your databases because the maintenance procedures need to be deployed to every single database.

 
Running database maintenance
Just like on-premise databases, Azure SQL databases also need to be serviced every now and then. To make sure I’m not reinventing the wheel I’ve re-used a solution that has proved its worth in practice: Ola Hallengren’s “SQL Server Maintenance Solution”.

But because every Azure SQL database is a contained datastore, I’ve made some minor changes to his solution. You can read more about that here.

 
Performance issues & alerting
One of the biggest differences for me between an on-premise database and an Azure SQL database is how I need to determine the cause of performance problems and solving them. When you’re working on a on-premise database it’s easy to just look at the CPU and RAM used by your instance, and watch the disk IO in the performance monitor in Windows. But you don’t have those in Azure of course. There you need to work with “DTU” percentages, and “Data IO” and “Log IO” counters. But DTU (Database Transaction Unit) makes monitoring performance a bit too abstract if you ask me. Just look at the explanation Microsoft gives us regarding DTU:

 

The Database Transaction Unit (DTU) is the unit of measure in SQL Database that represents the relative power of databases based on a real-world measure: the database transaction. We took a set of operations that are typical for an online transaction processing (OLTP) request, and then measured how many transactions could be completed per second under fully loaded conditions.

For example, a Premium P11 database with 1750 DTUs provides 350x more DTU compute power than a Basic database with 5 DTUs.

 
One thing I’m really happy with is that Adam Machanic published a new version of his sp_WhoIsActive for Azure a few months ago. This gives me the opportunity to quickly look at the results to find a cause of the performance issues.

 
When after a while you have a gut feeling about what the performance of your databases, you run into a new challenge with monitoring: the monitoring of DTU usage has a maximum retention of just 1 hour. This gives me another challenge, because I would like to see a longer retention. I definitely don’t want to over-scale my databases, so in some cases a long running process can trigger this alert. And when this happens during the day it’s annoying, but still okay. But it also woke me up in the middle of the night during an on-call rotation. This forced me to set some of the alerts on the maximum value of 45 minutes / 1 hour, and just disable others. In my opinion, this could use some work by Microsoft.

 
Conclusion
So in general I’m REALLY satisfied with the functionality that Azure SQL database provides us with, even though there’s always something to wish for of course. The move to Azure gives me as a DBA a new set of challenges and (in some cases) demands other/new skills, but it also allows me to do more than just manage a bunch of databases. I’m now also a part-time system administrator, network admin, (data)architect, developer, etc. This is something I personally really like, because it’s exactly that that allows me to broaden my horizon, and use a set of skills I didn’t for a while when working on on-premise databases.

So even though some people were afraid the DBA role would disappear because of the cloud-uprise, I’m convinced it will be here to stay except it will be a role with more skills needed than before, and I think that’s a good thing.

Advertisements

Invitation: T-SQL Tuesday #82 – To the cloud… And beyond!!!

T-SQL Tuesday is a recurring blog party, that is started by Adam Machanic (Blog | @AdamMachanic). Each month a blog will host the party, and everyone that want’s to can write a blog about a specific subject.

 
This month I’ll be the T-SQL Tuesday host, and I’m really honored! Thanks for inviting me to be the host Adam!

 
The topic
When Adam asked me if I wanted to host another T-SQL Tuesday, I immediately knew a topic I wanted to talk about: The cloud, and (if you want to) specifically about Azure SQL database.

Last time we blogged about the cloud was back in december of 2013, when Jorge Segarra hosted this monthly party. Since then, “the cloud” (to use that buzz-word again) has changed a lot, and I think the possibilities are endless nowadays.

The reason I would like to see you all blog about this topic, is that I’m working with Azure SQL databases a lot now since I switched jobs in December. Currently I’m working for a small start-up that has a cloud-first focus. This means the main (if possible) host for our (data)platform is Azure. And although hosting our platform in Azure makes it easier for us to scale parts of that platform, it also gives us new challenges to overcome. And I’m probably not the only one with that experience…

So with that said, I would like to give you the opportunity to blog about the cloud, in the broadest sense that you can imagine. I’m imagining really interesting blogs about migrating to the cloud, missing features in Azure SQL database, how you’re determining the cause of and solving performance problems, etc. Please surprise us all with your view on the cloud.

 
The rules
– Your post must be published between 00:00:00 UTC and 23:59:59 UTC on Tuesday 13th of September
– Include the T-SQL Tuesday logo in the top of your post, and your post must link back to this one (trackback and comments are moderated, so it might take some time before they’re visible) or tweet about it using the hashtag #TSQL2sDay
– If you like this, check Steve Jones’ (Blog | @way0utwest) blog post that contains the list of topics, and contact Adam Machanic (Blog | @AdamMachanic) if you’d like to host this party yourself

T-SQL Tuesday #48 – Cloud Atlas

T-SQL Tuesday is a recurring blog party, that is started by Adam Machanic (Blog | @AdamMachanic). Each month a blog will host the party, and everyone that want’s to can write a blog about a specific subject.

This month the subject is “Cloud Atlas”. If you want to read the opening post, please click the image below to go to the party-starter: Jorge Segarra (Blog | @SQLChicken).



 
In the last few years, “the cloud” has become more and more important in our lives. Not only in IT, or as a database- or data-professionals, but also in our personal lives. Take a look around you. How many people do you still see, carrying around a USB drive to store their data? Now do the same, and count the people that use a cloud solution for their data, like Dropbox, SkyDrive (if we are still allowed to call it that…), or Copy.com?

So everyone is storing their data in the cloud now. From personal information like a copy of a passport, to text files with peoples password lists. So without jumping to conclusions just yet, I guess we trust the companies that hold our data right…?

 
Trust
But now comes the hard (and controversial) part: we trust them with our personal data, but not our corporate data. It’s okay to store your passwords and private documents in the cloud, but it’s unthinkable that you store business data in the cloud!

So where is that distrust coming from? It probably has something to do with the whole NSA-thing. There, I said it! Without completely going off-topic, I would like to explain something about this statement.

My personal opinion is that people in the Netherlands are different from the rest of the world, when it comes to their privacy. They don’t care if the ISP is monitoring web traffic. They know it’s being monitored, but they accept that as a fact. When it comes to downloading games, music or movies, they think their entitled to that. But when it comes to government agencies monitoring the corporate data they put in the cloud, they draw the line.

 
Are you… the one…?
In the past few years, the discussion about on premise and off premise data intensified. People try to convince each other with arguments, and think the other is completely wrong.

A while ago, I encountered my first “cloud-company”. I’ve done some consulting for them, and they’ve set themselves the goal to move to the cloud within the next few years. The biggest advantages they see are costs, scalability and administration. And I fully agree with them.

 
Why use a cloud solution
Choosing a WASD (Windows Azure SQL Database) solution makes it easier to maintain your environment. You don’t have to monitor the hardware, and move to another server if your hardware fails or dies. This is all being taken care of by Microsoft.

Looking at the cost of a cloud solution is easy: it saves you money. When you run on premise servers, where you need a data center, electricity, maintenance team, etc. When you use a cloud solution, you only pay for the hardware you need. And if you’re done with it, you can just shut down the machine you were working on.

The same goes for scalability. For example, if you need to run a billing process, you could “spawn” twice as many cloud machines. This makes scalability a piece of cake. And again, when your done, just get rid of the machines you don’t use anymore. This makes it easier for companies to run big processes in a smaller amount of time.

 
Trying it out
The only time I’ve used WASD is on the machine that Jamie Thomson (Blog | @jamiet) made available to the SQL Family (read about it here). This was later taken over by Red-Gate, but I’m not sure this is still available.

But if you want to try it out, just create your own AdventureWorks on Azure. You can download the scripts here, and start your Azure trial here.