Share all details about your problem, including any error messages you may have received.
This is a followup to an earlier question/thread:
Amazon AWS has 2 different DB products that are supposed to be compatible with MySQL:
RDS MySQL
Aurora
From my reading, “Aurora” is an Amazon-implemented AWS-hosted database that is supposedly compatible with MySQL and Postgres, and it supposed to be more performant (plus has additional differences vs.) than the RDS MySQL (but possibly more expensive).
We are just starting to do some performance testing with Sailpoint with the RDS MySQL, and I am thinking maybe comparing to Sailpoint+Aurora vs. Sailpoint+RDS MySQL, so I am wondering if anyone has tried using Aurora with Sailpoint IdentityIQ?
If so, specifically, initially, I wanted to find out if there are any “gotchas” with integrating Aurora as the DB for Sailpoint?
I’ve done a bit of searching, and found almost nothing about this. I did find one post/thread here from several years ago, but it didn’t have much info in it.
We leverage AWS Aurora MySQL for our IIQ deployment. The additional capabilities are nice (replicas, automated backups, etc.). Latency is somewhat higher than you would experience with an on-prem DB, but I think that is going to be the case for any AWS DB. We’re using AWS’s custom JDBC driver and have had no issues. It allows you to define some failover parameters and other properties that the normal JDBC driver does not support.
Was there any discrepancies with schema, etc., e.g., running DDL with Aurora vs. MySQL?
Also have you all run any benchmarks with the Sailpoint Database Performance tester? FYI, we have some numbers from our tests with RDS MySQL:
IdentityIQ Database Performance Test starting
IdentityIQ Version: 8.4p1 xyz3e6f4783-20240325-035yyy
DB Performance Test Rule version: 20190827
Executed from Host: xyz9b6ea87ec
No schema issues whatsoever. Your DB performance tests were inline with what we experienced as well. Ultimately, any real performance issues we encountered were not due to the DB. Our DBAs and AWS support monitored our DB usage and did not find anything particularly troubling on the DB side.
My experience with RDS is that it does not meet the DB performance requirements. I have seen this affect application performance on very large systems. You can mitigate this somewhat with additional task servers to increase thread pool sizes, but the database issue remains. I also do not recommend MySQL for production deployments of any kind, due to its record and index size limitations. When building a companion sandbox on MySQL, I often have to change the size of the extended attribute fields to allow named columns to be created.
I also have seen some performance issues on systems with cloud server based databases (not cloud databases) but those are normally only on the 1K tests, the 4K and 8K tests are in limits. This is something that can be fixed. It’s normally because 8K is the packet size in the database configuration.
Steve - Sorry I have a question… FYI, the numbers that I posted were for RDS MySQL, not for Aurora DB. So when you said the tests were inline with what you saw, does that mean that the performance you saw with Aurora was similar to the performance with RDS MySQL?
As I mentioned in the OP this current thread was a followup to an earlier thread, where we were trying to change our original Sailpoint installation, which was using an MS SQL (running on an AWS Windows machine) to use the RDS MySQL. FYI, I think the impetus for that move was so we didn’t have to deploy a DB ourselves (but that is just a guess because I am fairly new to this team).
Given that we are kind of early in our dev cycle, what DB (either running on an AWS EC2 instance, or in RDS), working with Sailpoint, WOULD be recommended vis-a-vis performance?
For clients up to about 300,000 identities I normally recommend either cloud server based SQL Server 2019 or Oracle 19c. For more than 300,000 identities, I recommend Oracle. I did have a client seriously considering PostgreSQL but I haven’t had the opportunity to evaluate.
I know that this was not you reco, but after we discussed the test results with the RDS MySQL DB, we then reverted to the original configuration where we used MS SQL on an EC2 instance and ran the DB tests. Results of that test was that it was within the target numbers.
But, they prefer not using MS SQL because of licensing, so then we modified the configuration to use a MySQL DB on an EC2 RHEL8 instance, and DB test results of THAT configuration are kind of in-between the MS SQL test (fastest) and the RDS MySQL test (slowest).
I’m waiting to hear back/discuss what they want to do, but I was wondering if you could clarify what you meant exactly by “cloud server based SQL Server 2019”?
Also you had mentioned that adjustments could be made. Did you mean that there might be tweaks for the MySQL one EC2 configuration that might bring the performance in line with the MS SQL on EC2 configuration?
Also, FYI, my impression of the number of identities for this environment will not be huge… definitely less than 300K identities.
When I say “cloud based” I mean EC2 or EC2-like, there are multiple cloud infrastructure providers. I just don’t want to favor one provider over another. And SQL Server 2019 isn’t that the latest version? I really just mean use the latest certified version of any database.
The only tweaks I can suggest are to make block size 1K instead of 8K and honestly I don’t have any experience doing that, it’s just an “I think that might help” thought.
IF you are a smaller install and IF you don’t need a ton of named columns and IF you can up front set the extended attribute column widths to 100 instead of 450, then MySQL might be acceptable. It’s your decision.