┊・databases

[proud pond]

Just want to welcome those that are interested in this channel. By way of introduction, my name is Taylor Riggan and I'm a Sr. Graph Architect at AWS. I work on the Amazon Neptune service team. Neptune is AWS's managed graph database. Outside of Neptune I also spend time contributing to open source projects and communities such as Apache TinkerPop and various other graph libraries. I also have a lot of interest in MySQL and Postgres and spend a lot of time tinkering within those platforms. Prior to my time at AWS, I spent a number of years as a consultant (with a heavy focus on NetApp technologies) and performed a number of Oracle and SQL Server migrations/sizing exercises from other storage platforms to ONTAP-based arrays.

[cedar yarrow]

hi @proud pond I just happened across the discord from a youtube video from Nick Howell(aka DatacenterDude) and I am just learning all that NetApp ONTAP has to offer for DB engineers regarding IOPS/Latency/Backups/etc... What are the best resources to discover a Database focus on the information?

[proud pond]

@cedar yarrow - Likely the best thing to do is lookup the associated NetApp "Technical Report" for the database engine that you are looking to use with NetApp ONTAP. Technical Reports, or "TRs" as many NetAppers like to call them, are technical whitepapers that cover a lot of the best practices in a particular area. For example, here are a few key ones associated to using ONTAP with the more popular commercial and open source relational database engines:

• SQL Server w/ ONTAP: https://www.netapp.com/pdf.html?item=/media/8585-tr4590.pdf
• Oracle w/ ONTAP: https://www.netapp.com/pdf.html?item=/media/8744-tr3633pdf.pdf
• MySQL w/ ONTAP: https://www.netapp.com/pdf.html?item=/media/16423-tr-4722pdf.pdf
• PostgreSQL w/ ONTAP: https://www.netapp.com/pdf.html?item=/media/17140-tr4770pdf.pdf

Is there a particular database engine that you are primarily focused on?

[cedar yarrow]

https://tenor.com/view/frickin-awesome-amazing-gif-20449310

@proud pond , you ROCK! thanks so much for all this great info, now I know what I am doing for the next 12 hrs... haha.
I am interested in all of these plus all the next-gen databases like MongoDB, Cassandra, etc...

[swift hedge]

https://netflixtechblog.com/migrating-netflix-to-graphql-safely-8e1e4d4f1e72

Nice read!

[proud pond]

Probably one of the biggest pieces of confusion in the dev world today... GraphQL has little to do with databases. It isn't a "query language" for a database, but instead a replacement to using REST (which was a replacement to SOAP/XML and its precursors). Facebook built GraphQL to query their "graph" which was/is more of a logical construct of their social graph. GraphQL translates front-end calls to fetch data from various backend systems using the concept of a logical GraphQL schema and a series of "mutatations" and "resolvers". It sounds a lot like a database, but it doesn't necessarily care about the underlying data persistency/storage.

[cedar yarrow]

@proud pond I see you have migration of Oracle from other vendors experience, can you share any tips, tricks, white papers, etc?

[proud pond]

It's been a few years since I was doing those types of migrations. Just looking around for a good doc to provide you turned up this: https://www.netapp.com/media/19750-tr-4534.pdf

Most of the migrations that I have done in the past were just ASM-based migrations. So we just used ASM to provision new extents on the NetApp array and roll off of the old extents. There are some tradeoffs in this approach, as the back-out plan can be tedious. Some folks prefer doing full database copy migrations, but these will generally come with more downtime.

For any migrations, I would suggest some of the following tips (learning from my own battle scars):
a) Ensure you have proper connectivity from the host to the target NetApp array (ensure HA across switches, proper multi-pathing is being used, etc.). Create yourself a checklist of each of the configuration steps for each host along with a set of test to validate that HA and multi-pathing is working properly. Do all of this before even thinking about provisioning storage and attempting the migration. This can be more tedious with FC/FCoE with zoning/masking - keep a checklist.
b) Get baseline performance metrics from the old platform. Know when there are times of high I/O and if there are times where the database might be I/O-limited based off of limitations from the older array. This has bitten me in the past where an older storage array was not providing enough I/O to a database only to find out that the NetApp replacing it absolutely crushed it. You may not think this is a problem, but moving a bottleneck in a system from one location to another can have certain side-effects. Just understand the performance profile from the older system first to be on the lookup for these sorts of issues that might crop up post-migration.
c) Test, test, and re-test your migration workflow. Ensure you have a back-out plan and ensure that you can perform a back-out plan within your target migration window.

And most importantly, make sure you have backups, that those backups are good (and tested). And just for good measure, make sure you have a solid DR plan that works if things really go sideways.

[cedar yarrow]

@proud pond much appreciated. I like the ASM approach, but agree, the backout could be tedious... so as everything, it depends... tradeoffs, etc.
thanks for the lessons learned section and yes, moving the bottleneck can have negative effects as well...

[cedar yarrow]

Thanks for all the help @proud pond I didn’t get the job with NetApp.

[woeful dew]

hello there, just wondered if there are magazines mainly about netapp? I mean ofc I can Google all day to find articles,but having a mag that puts them together with all the charms of side topics would have its own charm.

[spare dagger]

New to NetApp discord, past rare user of Discord in general.
I have some very specific questions about OnTap performance in an active-active HA cluster and partitioned databases. Is there a preferred channel for this deep dive conversation? I have some interesting observations and would like to confirm or refute some assumptions!

[graceful pendant]

Welcome! If you would, let’s use #1062049169520476220 for the discussion!

[median acorn]

Here's a question for Oracle DBA's. For migrating 19c stand-alone instances on NFS mounts on NetApp filers, I want to use these steps to migrate from RHEL7 to RHEL8. Are these steps sufficient?

1. Create new RHEL8 servers and install new Oracle Homes at same version as source RHEL7 servers
2. Copy folders and files(https://dohdatabase.com/2023/05/30/files-to-move-during-oracle-database-out-of-place-patching/) from source $ORACLE_HOMEs to new $ORACLE_HOMEs and update /etc/oratab, listener.ora, sqlnet.ora, etc... like we would do if we were performing out-of-place patching
3. Shutdown databases, unmount NFS mounts from RHEL7 servers, and mount to RHEL8 servers
4. Startup databases on RHEL8 servers and run @?/rdbms/admin/utlfixdirs.sql to update internal directory structures for any new paths

[swift hedge]

We used to do a quicker test/dev way, but this was a long time ago where we would restore the latest snapshot of production, mount that, and then run a SQL script that would replay the online redo logs to bring it up to-date… Whole process took about five minutes

But that was with Oracle 10 G and 11 and I’m not sure what’s changed since then

We also automated this using snap manager for Oracle, so if you have snap center with the Oracle plug-in, you could probably make this a very easy and fast transition

[misty gull]

Hey, does anyone know about CRUD Operation in SQL?

[jovial sierra]

yep, they are INSERT INTO, SELECT FROM, UPDATE and DELETE FROM (respectively) in SQL

[misty gull]

thanks

[swift hedge]

https://blogs.oracle.com/ebsandoraclecloud/post/ebs-cloud-manager-24111-now-available

In this release, we have switched from Terraform to OpenTofu due to forthcoming Terraform licensing changes. You must therefore upgrade your Cloud Manager to 24.1.1.1 by June 30, 2024 at the latest.

Heads up, Oracle customers… this is pretty big news

[misty gull]

Hey, i am working on a website which should be like this -

A website which will be deployed online, this is a website related to a movie, in the landing page of a website there should be a question "which content you want next", the options should be teaser, song, trailer, people should vote for it and after voting it should show percentage and one more thing even if 5000 people vote at same time it should count them, how can i do this project, that 10k simultaneous voting is main concern think about that also. also one person should from one device can only one votes, you can think off how you can do that, check if my idea works, device mac id or some id you can track and restrict him from second vote in different browser and all, and also after he casts vote the button should disable, even if he click it should not enter the database for counting, this is so accurate project so every vote is important. Essentially, duplicate votes from same person should not be there.

What tech stack should I use?

Any help would be appreaciated ! Thank you

[proud pond]

I'm completely biased here (I work for AWS), but if I had to choose an architecture, I would have all of the votes being sent to an SQS queue. Standard SQS queues have a nearly unlimited throughput. You can just embed the calls to SQS into the browser-side code and have a message sent to the queue from the browser. You can then have a consumer reading off of the queue and processing the messages in batches (reduces the amount of hardware you need on the backend). The backend consumer can also dedupe the votes as they are being processed (which you'll want to do anyways, as Standard Queues are "at least one-time" delivery for all messages.

Having a queue between your clients and your database also allows you to change database architecture later if you want. You can move from MySQL/PostgreSQL to something like DynamoDB/Cassandra/NoSQL.

[fresh pollen]

To create a case, visit the NetApp Support Site
👉 https://mysupport.netapp.com 👈
Find more support options in #┊・community-support.

[plain heath]

Evening folks. Just wondering if anyone here has experienced with using NetApp replication for SQL always on FCI (not availability groups) - specifically wondering how the WSCF architecture looks. As FCI requires shared storage how does this work across a multi site setup (DR purposes not HA)

[opal flint]

I ran replication for sql FCI.
We didn't use multi-site though as it would constantly complain about slow access times to the storage.

[plain heath]

Yeah - if I add a node at another site and join it to the FCI it must have access to the shared storage at primary site, which would be over the s2s VPN - unless there’s a way round this.

[plain heath]

Can I ask why did you run replication if both nodes were on the same site?

[opal flint]

different locations in the building. different power sources and external connections

[plain heath]

Gotcha thanks