What is our primary use case?
We are using AWS Database Migration Service for our data migration from our RDS, Oracle database from on-prem or cloud to moving this data into AWS ecosystem, S3, and then into Snowflake. We are using AWS Database Migration Service, and now we are on a journey of onboarding the stream platform.
What is most valuable?
The cost effectiveness of AWS Database Migration Service is good as it is quite cost effective. The scalability option is another valuable feature because AWS provides its own compute behind it, so I can scale up and scale down at any given point.
The security features of AWS Database Migration Service, such as encryption and IAM integration, are good.
What needs improvement?
We have the continuous data integration feature, but it does not work very well in our given ecosystem because we have huge data volume, and it reads from archive log rather than the redo sequence. That is where it lacks, so we are not able to do continuous streaming data injection.
The ability to handle heterogeneous migrations for our organization is adequate. I do not see a major value add, or see it as a marquee feature. It is a good tool, but we do not have that sort of uses requirement. In my 10 plus years of experience, I do not see that kind of requirement where I have to replicate it in real time.
I leverage AWS CloudWatch for addressing migration issues. The major challenges come across CDC. We have created latency filter and everything, but there is no way to control it. Due to the nature of our business, the transactions increase during month end and year end. My transactions are supposed to increase, my workloads on my core system increases, but my DMS is not able to keep up with them, even if I increase it to a higher side.
Improvements could be made in AWS Database Migration Service. Because of that, I am moving out of DMS now to a STRIIM solution. DMS works within AWS ecosystem, but they also have to look for third party solutions. Now Snowflake is a bigger player, or Databricks. Something that could be possible as a connector, whether paid or free, but giving as a connector to that ecosystem as well.
For how long have I used the solution?
We have been using the solution for five years.
What was my experience with deployment of the solution?
The installation and deployment of AWS Database Migration Service is straightforward. If I am able to predict some loads, I can increase the capacity, spin up a new DMS instance just for my one-time load activity when onboarding a new system. If I do not want to have an impact on existing pipeline, I can onboard a new DMS instance for one-time activity, or DMS serverless, move ahead, and then again move back to my existing installation.
What do I think about the stability of the solution?
AWS Database Migration Service is easy to maintain. The only concern is that the patching window is huge.
The service is easy to maintain with one person automatically. My infrastructure team handles this, and we have taken it as automatic. For DMS version upgrades, we schedule downtime during business hours so that midnight workloads are not interrupted and morning business can run smoothly. This is a quarterly activity, not continuous. Minor version upgrades are handled by DMS automatically. For major version upgrades, we need downtime, but it is very manageable and not a bottleneck.
What do I think about the scalability of the solution?
While scalability is good, latency exists due to our business nature, and AWS Database Migration Service or AWS is not able to help in this regard. I have raised many cases and spoken to AWS SMEs and expert solution architects, but it remains limited.
How are customer service and support?
I am happy with the technical support from AWS, but the product enhancement that is supposed to happen is not there. On a scale of one to ten points, I would rate tech support as seven.
How would you rate customer service and support?
What was our ROI?
AWS Database Migration Service has definitely helped save money, but not time.
Regarding money savings, I can specify savings of around 40 to 60%. I have another benchmark as an ETL, ELT tool. If you take a legacy ETL tool within any data enterprise ecosystem, they will have Informatica, ODI, but they do not have a big data connector to connect to S3, Snowflake, or Redshift directly. I have to use Glue, PySpark, Python, or Airflow kind of custom made connector, but I have to give them heavy compute. Compared to DMS, those are expensive solutions. I am able to save around 40 to 60%. Otherwise, I would have to migrate those pipelines to Glue because that is the only ETL option in my arsenal. If I had Informatica cloud or Talend as a solution, I would have gone through that particular route. Here it is saving me around 40 to 60%, but for streaming solutions, I have to go for an expensive situation.
What other advice do I have?
We required the AWS Schema Conversion Tool in very smaller capability. We have an enterprise data lake which acts as a landing layer. We do not require it in a very high context because schema conversion would be from one database to another database. We do not do that kind of heavy migrations. First we land everything within our AWS ecosystem, then move it into Snowflake.
AWS Database Migration Service is affordable and cheaper. However, it is not able to solve our critical needs.
If anyone has questions or comments about my review, they can reach me by email.
I rate AWS Database Migration Service as nine out of ten.
Which deployment model are you using for this solution?
Private Cloud
If public cloud, private cloud, or hybrid cloud, which cloud provider do you use?
Amazon Web Services (AWS)