Azure vs AWS for databases

  • David Webb-CDS

    SSCoach

    Points: 17338

    Folks,

    I know this is a pretty broad topic, but I'm looking for anything that anyone can provide on analysis you may have done on price/performance metrics between SQL Server on Azure and Postgres on AWS. We have a couple of camps arguing that SQL Server on Azure performs better and that Postgres on AWS is a lot cheaper. Any useful input anyone could provide will be welcome. The company currently runs SQL Server only and only on-prem, so this is a big leap no matter which way they go.


    And then again, I might be wrong ...
    David Webb

  • Alan Burstein

    SSC Guru

    Points: 61039

    I'm no expert at either but have worked with both. I'm going to assume that by, "Postgres on AWS" you are talking about Redshift.

    My experience has been that a beefy AWS Redshift instance performs significantly better on huge datasets than any comparable Microsoft solution. Many of my colleagues (BI Developers and Architects) would tell you the same. If you have worked on predominately Microsoft products, however, you will find the AWS web interface to be clunky and unpleasant. It took me nearly all day once to figure out how to move a small data set from an S3 repository into a Redshift data warehouse.

    "I cant stress enough the importance of switching from a sequential files mindset to set-based thinking. After you make the switch, you can spend your time tuning and optimizing your queries instead of maintaining lengthy, poor-performing code."

    -- Itzik Ben-Gan 2001

  • Eric M Russell

    SSC Guru

    Points: 125020

    What are primary motivations for migrating to a new database platform?

    For example: line of business transactional performance, analytical and reporting performance, flexible storage and lower cost of ownership?

    "Do not seek to follow in the footsteps of the wise. Instead, seek what they sought." - Matsuo Basho

  • David Webb-CDS

    SSCoach

    Points: 17338

    The company wants to add large amounts of new data into their analytics base and also move infrastructure off their current hosted environment. TCO is always an issue, but scalability and flexibility trump short-term cost savings.


    And then again, I might be wrong ...
    David Webb

  • Steve Jones - SSC Editor

    SSC Guru

    Points: 715841

    Are you looking for a platform or hosted VMs with a database software running?

    I'm wondering how you want to compare SQL Server v PostgreSQL as well. Aren't there different costs for development or admin across these? If you have staff and don't care, PostGreSQL should be less licensing, so lower cost.

    Both can work well, or work poorly. You need to consider the cloud platform and how latency affects you. AWS is more mature, and some solid systems run on it every day (Netflix, etc) and it seems to have fewer outages. I find Azure to work great sometimes and be flaky others, but I've run across customers that love it and find it to work very well. I think if you pay for performance, then at larger sizes of VMs/DTUs/whatever, it can work well.

    The MS platform has lots of tooling integration. PostgreSQL is mature, but I think the AWS tools can seem raw for an MS stack developer. Again, staff costs are something to consider.

    Cloud pricing seems to vary dramatically, and changes quarterly. I think AWS is less expensive, but that can change.

    It's a hard choice. I think with either of these, you have lots of flexibility, but you'll burn time/salary to do whatever you want. Ideally, you want to fit within their paradigm closely to limit the development costs and efforts.

Viewing 5 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic. Login to reply