1. Comparison of AWS Databases
All right, in this lecture, let’s compare different AWS database solutions. So this will help you to choose the right database for the right job. So here’s the comparison of some of the AWS databases. All right, so we have a mix of relational databases, data warehouses, no SQL databases, as well as the Graph database. So first, let’s talk about the relational databases. So, for relational databases, we have RDS and Aura. Relational databases are meant for structured data. We know that workload is OLTP, so we have relational data with OLTP or online transactional processing workloads. Or you can also use them for simple OLAP use cases.
You can use them for analytics, but not for complex analytical use cases. For complex OLAP, you would go for a data warehouse like Redshift. So if you compare RDS and Aura, RDS is meant for low TB range databases, so smaller databases, whereas Aura can support mid range or mid TB range databases. When we look at the performance, then RDS is meant for medium to high throughput and low latency requirements, whereas Aura can give you high throughput along with low latency. And again, operational overhead is moderate with RDS, and you have low to moderate operational overhead with Aurora.
Then again, we have Redshift, which is meant for OLAP workloads or for data warehouses. It’s also a relational database, and it can support structured as well as semi structured data. And this is typically used when you have very large data set, like in PB or Petabyte range. If you have smaller data, you can definitely consider using Athena. And when we talk about data warehouses, latency is not really a major requirement. Redshift provides a mid to high latency, and operational overhead is moderate. Then we have DynamoDB, which supports semi structured data. This is a non relational key value store. You use it for OLTP workloads.
You cannot use DynamoDB for analytical workloads, right? Remember that. And this can also be used as a document store. So we have two document databases. One is DynamoDB and document. DB. Now, Document DB is not listed here, but you can definitely use Document DB to store your JSON documents. And you can also use DynamoDB for the same purpose. Now, DynamoDB supports high TB range, so it can store a huge amount of data higher than Aurora as well. And you would typically choose DynamoDB when you need ultra high throughput with ultra low latency, right? So ultra high throughput is provided by DynamoDB, and DynamoDB latency is in single digit milliseconds.
And if you use DAX or DynamoDB accelerator, you will get ultra low latency or microsecond latency. Also, this is a serverless database, so operational overhead is very, very low. Then we have Elastic cache. It’s a caching service. You use it for caching purposes, so you can use ElastiCache for microsecond latency. So this again is ultra low latency. It provides you with a high throughput, but not as much as DynamoDB. All right? So ElastiCache can store semi structured data as well as unstructured Data. And ElastiCache again is a Key Value store. Just like DynamoDB. You would use it for non relational data? For in memory caching.
Now, when I say non relational, it does not mean that you can’t use elastic cache on top of RDS or Aura. You can definitely do that. But remember, even if you store relational data in Elastic cache, it’s going to be stored in key value format. Now, ElastiCache database sizes are in low TB range. Okay. It’s typically used for smaller databases. So if you have larger databases, then DynamoDB is something you can consider using our DynamoDB along with tax if you want. Ultra low latency. So, as I mentioned before, ElastiCache gives you microsecond latency, which is ultra low latency. And again, it Gives You a high throughput. If you need even more throughput, go for DynamoDB.
Again. ElastiCache is simple to operate, so operational overhead is low. And finally we have Neptune, which is a graph database. It’s used for graph workloads or highly connected graph data sets. Again. It supports a mid DB range. Database sizes. The sizes are comparable to what we have with Aura because the architecture of Neptune and Aura is same. And Neptune gives you high throughput with low latency. Again with Neptune. The operational overhead is low. So that was a quick comparison of AWS databases, and hopefully this should give you some cues on choosing the right database for the right job. All right. So with that, we come to the end of this section. Let’s continue to the next one.
