Elasticache

Question 1

What are some key differences between Reddis and Memcached?

Answer 1

Memcached: Key/Value pairs. Multithreaded. Good for simpler cache use cases.

Reddis: Complex data types and structures. Can be persisted to disk, can sort and rank data.

Question 2

What are the 2 engines used by elasticache

Answer 2

Reddis and Memcached

Question 3

Can a write through cache be stale?

Answer 3

No. With a write through cache as data is written to the database, it is also written to the cache - the cache is therefore always current

Question 4

Is Elasticache multi-AZ for both Reddis and Memcached?

Answer 4

yes to both

Question 5

I need a multithreaded in memory cache - would I use Memcached or Reddis?

Answer 5

Memcached is multi threaded

Question 6

Which cacheing engine supports complex datatypes, ket store persistence and read replicas for read intensive applications?

Answer 6

Reddis

Question 7

Would we expect reads in an application using write through cache to be faster or slower than those for lazy loading?

Answer 7

Faster - as cache is kept up to date

Question 8

What is a DISADVANTAGE of using lazy loading (aka cache aside/lazy population)

Answer 8

Data can become stale when using a lazy loading strategy. As lazy loading is read based, if it finds a hit in the cache it will use that - however if the underlying data in the database has changed, this data will be out of date.

Also, there is a read penalty that is incurred with lazy loading.

Question 9

For lazy loading, how many round trips occur and to where when there is a cache miss?

Answer 9

3. First round trip is to the cache to return the miss. Second round trip is to the database to select the data. Third round trip is back to the cache to update it.

Question 10

Describe this code and the caching strategy it reflects:

def get_user(user_id)
 record=cache.get(user_id)
 if record is None:
    record=db.query("select * from users where Id=?", 
                                  user_id )
    cache.set(user_id, record)
    return record
else:
    return record

user=get_user(12)

Answer 10

This represents a lazy loading cache strategy. Here, we attempt to retrieve the user_id from the cache. If its a miss (record is None) we execute a select against the db and write the result back to the cache and then return it from the function

Question 11

Does memcached allow for data persistence?

Answer 11

No, it is in memory only. If the node goes down the data is lost

Question 12

What scaling actions will you need to take if you are seeing a large number of cache evictions due to memory constraints?

Answer 12

Its likely you will need to scale up or scale out.

Question 13

Who is responsible for a cache invalidation strategy?

Answer 13

As the developer - you are.

Question 14

When starting a memcached elasticache cluster, will the cache be empty. What about for reddis?

Answer 14

Memcached will always start empty. Reddis can be restored from backup

Question 15

Which elasticache engine is multi-threaded

Answer 15

Memcached

Question 16

Both Elasticache and DynamoDB can be used to store session data for a web application, and both use key value pairs. Given this, why would you use one over the other?

Answer 16

you would use Elasticache if you specifically needed an in memory datastore.

You would use DynamoDB if you needed a SERVERLESS solution that can scale automatically.

Question 17

Which elasticache engine supports read replicas and which supports sharding?

Answer 17

Reddis supports read replicas, Memcached supports sharding

Question 18

Name 2 use cases for elasticache

Answer 18

DB offload (for reads), Session management (for stateless applications)

Question 19

What are TWO disadvantages of a write through cache?

Answer 19

1. While data cannot be stale - it can be missing until it is added or updated. This can be mitigated by using write through cache in conjunction with lazy loading
2. Cache Churn - a lot of data will be written to cache, but it may never be read. This can be a problem if you have a small cache.

Question 20

How many memcached nodes can you have per cluster. How many for Reddis? Which supports replication groups?

Answer 20

20 for memcached, 1 for reddis. Reddis clusters can be grouped into replication groups

Question 21

What are the 2 key caching strategies you can use

Answer 21

Lazy Loading and Write Through

Question 22

There are 3 ways to evict an item from cache. What are they?

Answer 22

Specifically Delete an Item
Eviction due to high memory utilisation
TTL

Question 23

What sort of penalty does lazy loading incur - read or write?

Answer 23

Read due the 3 round trips needed

Question 24

Name two ADVANTAGES of lazy loading

Answer 24

1. You only have required data in the cache so it isn’t filled with un-used data. This is because the cache is populated on a MISS
2. Node failures are not fatal. If a node goes down, there will increased latency as the nodes are “warmed” (i.e. caches are repopulated)

Question 25

Describe this code and the caching strategy it reflects:

def save_user(user_id, values)
 record=db.query("update users ... where id=?", user_id, 
                               values)
 cache.set(user_id, record)
 return record

user=save_user(17, (“name”: “Scott Tiger”))

Answer 25

This reflects a write through cacheing strategy. As soon as a record is written to the database, it is also written to the cache

Question 26

You need to implement a caching solution within your application that allows allows for backup and restore. Would you use memcached or reddis?

Answer 26

Reddis supports backup and restore

Question 27

If I wanted a caching engine which allowed for pub/sub capability - which would I use, memcached or redis?

Answer 27

Redis.

Question 28

Which caching engine allows for horizontal scalability and why?

Answer 28

memcached as it is multithreaded and allows for 20 nodes per cluster.

Question 29

I need a caching engine that allows for multi-az operation for fail over. Which would I use?

Answer 29

Redis.