By using AWS re:Post, you agree to the Terms of Use
/[ElastiCache] How to figure out the required nodes and instance in ElastiCache/

[ElastiCache] How to figure out the required nodes and instance in ElastiCache

0

Hi,

I am trying to figure out the required nodes and instance type in ElastiCache.

It is because I am trying to reduce the cost of following result by replacing DynamoDB to ElastiCache. I haven't decided should it be Memcached or Redis. However, the DB can be volatile, so I am considering Memcached for upscaling from Auto Detect.

The following is is the price calculation and needed data for cost calculation for your reference. Could you please guide me on how to get node number and instance type for ElastiCache?

In short, it's about 0.5GB data and there are high frequency of reads and writes. It looks like one node of ElastiCache/Memcached with high tier instance type would suffice, but I would like your advice.

Thank you,


https://calculator.aws/#/estimate?id=7ab704749c01bf87fbd852425e35d45f7a825d5e

Total: 31,195.07 USD

Devices: 60000

Average message size: 5800 bytes

Message per min while on: 6

Message per month while on: 13024606= 259200

Assume users only using IoT devices 10% of all time

Message per month: 25920

Total message instances: 60000 * 25920 = 1555.2 M

Assuming only 5% are different data = 77.76 M

DynamoDB item size: 6K bytes (including some meta data)

Total hot data size = 6KB * 60000 = 352MB

Assuming 10ms for each Lambda processing.

128MB used for Lambda

Cold data amount = 77.76 M * 5800 B = 451 GB

Compression assumption 10% = 45.1 TB

Infrequent access percent 90%

Objects per month (100 instances to 1 object): .7776 M

Read access of S3: 1% of write: 7.7760 K

asked a year ago21 views
1 Answers
0
Accepted Answer

I would suggest building the Cache on the Redis engine. ElastiCache redis provides greater resilience(multi-az), scale up/down, and security (encryption in transit). This article provides a great perspective on how to size your cluster (https://aws.amazon.com/blogs/database/five-workload-characteristics-to-consider-when-right-sizing-amazon-elasticache-redis-clusters/) and I would consider the Graviton based cache nodes (m/r6g).

MODERATOR
answered a year ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions