Monday, May 29, 2017

To AWS or not to AWS

Everyone is talking about moving to cloud these days for a good reason. Who wouldnt want to be serverless and having autoscaling servers , lowering the cost of all the licencing  and maintenance. And in fact not doing any maintenance and delegating that task to a PaaS/IaaS company.

AWS is the leading PaaS in the market offering over 60+ services along with it.
AWS platform includes  own queue service, own RDBM, own NoSQL db, even their own source control applications, project management tools .. .
And if you ask an Amazon rep, they would suggest to move alltogether to these services rather than just using AWS as a PaaS or IaaS.
But are all of the AWS services real good?
Why would I prefer DynamoDB over Mongo or any other NoSQL?
Why would I prefer TFS to CodeStar or whatever AWS Project management CI tool ?

I wanted to do a small research on these and here is a brief summary.
Lets look at those "services":

First of all all these services are pretty new, and everyone knows it takes many years to have a software mature , and at the moment none of these services are the leading software in the respective arena,
Aurora is not #1 in RDBM, DynamoDB is not # 1 in NoSQL arena, neither Redshift in the OLAP world.

And the fact is many of these AWS applications are modified versions of some Open Source software:
Aurora - Fork of MySQL(Amazon claims that Aurora is 5 x faster than MySQL)
Redshift - Fork of Postgre 8
ElasticCache- Using Redis and Memcached nothing really new
When you look at these services with their unmodified original versions yes many of them are quite popular.


The race to to dominate the Cloud market is going super fast and obviously Amazon doesnt want to reinvent the wheel in that race so they are using the available Open Source software which is pretty realistic.

Yet I find it very unrealistic to switch to entire AWS services altogether at the moment because of the fact that it takes many years for any software to mature.

And I also think Amazon may be thin spreading by trying to create a service for anything, instead they can buy market leaders , like TeamCity for CI or JIRA .

Edit: Dynamodb is very limited compared to other nosql databases, the query syntax overcomplicated

No comments:

Post a Comment