AWS Certified Solutions Architect Associate exam: numbers and key things to remember
This is a compilation of numbers and key things you need to remember as one of the preparation going into the exam. Use it as quick reference guide throughout your exam preparation time. Review it the night before your exam. To top it off, review it again one last time a few hours before you take your exam exam. That was how I did it and for me, it was helpful, and I certainly hope it will be useful for you too.
The AWS Certified Solutions Architect — Associate covers quite a broad spectrum of topics. As of July 2021, the SAA-C02 exam guide lists a total of 63 AWS services and features that an exam candidate needs to know as these are all covered in the exam.
My guess would be that it’s quite rare for anyone taking the solutions architect associate exam to have worked with all the 63 services and features, especially one who remembers all the service limits across all those. This is what the list is for.
Now, one might ask, does the AWS SAA-C02 exam really question you on these numbers? No not directly, they don’t ask you questions like “what is the maximum connection draining time you can set on an ELB?” Rather, they are typically baked into a scenario that you can answer correctly only if you know the service limitations and restrictions.
For example, there can be a question that asks you to choose which service to use for a compute workload that typically takes 20 minutes to complete, and they present you with Lambda and Fargate as the choices. In this scenario, you need to know the maximum execution time for Lambda, which is 15 minutes, to determine that it is not the right choice of technology to use.
This was the list that I personally use when I prepared for the exam, so it was in a way one of my personal study notes. It is by no means intended to cover everything, because the full list is all in AWS documentation which is very large. Here it is:
TTL for API caching: default is 300s. Allowable range is 0 to 3600s.
Cannot create stack across region, but can use StackSet to create, update, or delete stacks across multiple regions or even across AWS accounts.
1GB size limit for put/post, if dealing with bigger put/post requests use S3 transfer acceleration instead.
- Maximum item size is 400KB
- 25 unique items per transaction
- Maximum 4MB per transaction
- gp2 has 16,000 iops max. Size can range from 1GiB to 15 TiB.
- io1 has 64,000 iops max
- io1 or io2 type EBS volumes can be shared on up to 16 Nitro EC2 instances
- Spread placement group: max 7 instance per AZ
- Partition placement group: max 7 partition per AZ
- A partition placement group with Dedicated Instances can have a maximum of 2 partitions
- Cluster placement group: 1AZ only
- Public IP address change after hibernate or stop, retained if reboot
- Private IP address stays the same all the time
- Region bound
- Performance mode options: General purpose or Max IO
- Throughput mode options: Bursting Throughput or Provisioned Throughput
- Idle timeout: 60s default. Allowable range is 1s to 3600s
- Connection draining: 300s default. Allowable range is 1s to 3600s
Kinesis Data Streams
- Retention: 24h by default, but can be configured up to 7 days
- Maximum size is 1MB per record
- Maximum 1000 PUT per second per shard
- Maximum 2MB per second per shard for consumption
- The total unzipped size of the function and all layers can’t exceed the unzipped deployment package size limit of 250 MB
- Can have up to 5 layers
- Maximum execution time is15 mins
- Max memory 3008 MB, mininum is 128MB
- Multi thread only start if you allocate 1536MB of memory of above
- Typically other services can trigger Lambda (push to Lambda model), except for DynamoDB streams & Kinesis. With DynamoDB Streams and Kinesis, you have to configure Lambda to poll the stream (pull from Lambda model)
- Multi AZ read replica is only for mySQL
- No Read replica functionality available for MS SQL
- Auto backup read replica or enable writes only for mySQL and MariaDB
- Snapshots are not auto deleted on S3
Memorise this diagram well:
- Transition down only for objects that have been in previous tier for minimum of 30 days and minimum of128KB in size
- Glacier retrieval times: Expedited 1–5min, standard 3–5hrs, bulk 5–12hrs
- Deep archive retrieval times: standard 12hr, bulk 48hr
- Deep archive: minimum storage duration is 189 days
- Snowball: 80 TB
- Snowcone: 8TB
- Snowmobile is for 10PB or more
Supported target transports: HTTP/HTTPS, Email/Email-JSON, SQS and SMS
- Queue name maximum length is 80 chars
- FIFO with no batching: up to 300 msg per second without high throughput mode
- FIFO batching: up to 3000 msg per second without high throughput mode
- No api limit for standard queue
- Retention: 4 days default, configurable from 1min to 14 days
- Delay: 0 by default, configurable from 0 to 15 min
- Message timer: 0 by default, configurable from 0 to 15 min
- Visibility timeout: 30s by default, configurable from 0 to 12hrs
- Message maximum size is 256kb
- Long polling (receiveMessageWaitTime) maximum is 20s
- Standard queue can have up to 120,000 in flight messages
- FIFO queue can have up to 20,000 in flight messages
- NAT Gateway is at AZ level (always in public subnet), 45 Gps bandwidth limit
- NACL is at subnet level
- NACL default rules: allow all in/out, custom one deny all in/out
- Default & custom security group: allow all outbound
- Default security group allows inbound from itself
- Custom security group has no inbound rules (i.e. implicit deny all)
Best of luck for your exam!