Can you connect to Amazon ElastiСache Redis outside of Amazon?

Question

I'm able to connect to an ElastiCache Redis instance in a VPC from EC2 instances. But I would like to know if there is a way to connect to an ElastiCache Redis node outside of Amazon EC2 instances, such as from my local dev setup or VPS instances provided by other vendors.

Currently when trying from my local set up:

redis-cli -h my-node-endpoint -p 6379

I only get a timeout after some time.

Avenue

Answer 1

Update 2018

The previous answer was accurate when written, however it is now possible with some configuration to access redis cache from outside using the directions according to Accessing ElastiCache Resources from Outside AWS

---

Old Answer

No, you can't without resorting to 'tricks' such as a tunnel, which maybe OK for testing but will kill any real benefit of using a super-fast cache with the added latency/overhead.

The Old FAQ under How is using Amazon ElastiCache inside a VPC different from using it outside?:

An Amazon ElastiCache Cluster, inside or outside a VPC, is never allowed to be accessed from the Internet

However, this language has been removed in the current faq

Dulcle

Is this still the case? The docs no longer say this - they claim redis is governed by standard security group policies, but I still can't get access to my redis node despite that. Strike that. Ref just moved: Amazon ElastiCache Nodes, deployed within a VPC, can never be accessed from the Internet or from EC2 Instances outside the VPC. – Becalmed

I feel that 'kill' is a bit strong. For instance we get no appreciable performance hit when running our apps outside of AWS (via such a tunnel). The tunnel's overheads are minuscule compared to DB operations, browser load, disk I/O and so on. – Admass

docs.aws.amazon.com/AmazonElastiCache/latest/UserGuide/… – Aqueous

Answer 2

SSH port forwarding should do the trick. Try running this from you client.

ssh -f -N -L 6379:<your redis node endpoint>:6379 <your EC2 node that you use to connect to redis>

Then from your client

redis-cli -h 127.0.0.1 -p 6379

Please note that default port for redis is 6379 not 6739. And also make sure you allow the security group of the EC2 node that you are using to connect to your redis instance into your Cache security group.

Also, AWS now supports accessing your cluster more info here

Update 04/13/2024:

Many folks are running Kubernetes today. It's a very typical scenario for folks to have services running in Kubernetes accessing ElasticCache Redis.

So there is a way to do this (test your redis connection locally through Kubernetes) using the kubectl ssh jump plugin.

Follow the installation instructions. Then see case 2 here.

For example:

kubectl ssh-jump sshjump \
  -i ~/.ssh/id_rsa_k8s -p ~/.ssh/id_rsa_k8s.pub \
  -a "-L 6379:<your redis node endpoint>:6379"

and then from your client:

redis-cli -h 127.0.0.1 -p 6379

Bootie

Thanks for pointing out the port, just a typo. So, are you saying that SSH tunneling through EC2 is the only way to gain access to an elasticache node outside Amazon? Thanks, – Avenue

That's correct just like @E.J.Brennan mentioned in the other answer. – Bootie

How we can revoke ssh port forwarding...? – Townsman

you can kill the ssh process. On Linux: kill -9 <pid> – Bootie

Note that the -f option makes the ssh command execute as a process in the background, which may not be what you want. Use ps -ef | grep ssh to find the process_id (2nd column) so you can kill it with kill <process_id> – Pickett

Just FYI. Answer updated with specific Kubernetes use case. – Bootie

Answer 3

These answers are out of date.

You can access elastic-cache outside of AWS by following these steps:

1. Create a NAT instance in the same VPC as your cache cluster but in a public subnet.
2. Create security group rules for the cache cluster and NAT instance.
3. Validate the rules.
4. Add an iptables rule to the NAT instance.
5. Confirm that the trusted client is able to connect to the cluster.
6. Save the iptables configuration.

For a more detailed description see the aws guide:

https://docs.aws.amazon.com/AmazonElastiCache/latest/red-ug/accessing-elasticache.html#access-from-outside-aws

Intemperance

I don't want a NAT instance, I want to check on it for a minute. Rico's answer is exactly what I wanted. – Dope

how would the NAT allow access from the outside? the documentation specifies the following "Opening up the ElastiCache cluster to 0.0.0.0/0 does not expose the cluster to the Internet because it has no public IP address and therefore cannot be accessed from outside the VPC" Is there a way to allow this via NAT?? – Teahan

Answer 4

Not so old question, I ran to the same issue myself and solved it:

Sometimes, for developing reasons you need to access from outside (to avoid multi-deployments just for a simple bug-fix maybe?)

Amazon have published a new guide that uses the EC2 as proxies for the outside world:

https://docs.aws.amazon.com/AmazonElastiCache/latest/red-ug/accessing-elasticache.html#access-from-outside-aws

Good luck!

Opponent

For reference the approach Amazon mentions is a NAT instance. – Frangipane

FYI, from the docs: "This approach should be used for testing and development purposes only. It is not recommended for production use" – Shirashirah

Yes, that's true @jasonjonesutah I have also mentioned this in my answer. A very bad idea for production but excellent for development. – Opponent

Answer 5

BTW if anyone wants a windows EC2 solution, try these at the DOS prompt (on said windows EC2 machine):

To Add port-forwarding

C:\Users\Administrator>netsh interface portproxy add v4tov4 listenport=6379 listenaddress=10.xxx.64.xxx connectport=6379 connectaddress=xxx.xxxxxx.ng.0001.use1.cache.amazonaws.com

To list port-forwarded ports

C:\Users\Administrator>netsh interface portproxy show all

Listen on ipv4: Connect to ipv4:

Address Port Address Port

---

10.xxx.128.xxx 6379 xxx.xxxxx.ng.0001.use1.cache.amazonaws.com 6379

To remove port-forwarding

C:\Users\Administrator>netsh interface portproxy delete v4tov4 listenport=6379 listenaddress=10.xxx.128.xxx

Admass

Answer 6

We are using HAProxy as a reserved proxy server.

Your system outside AWS ---> Internet --> HAProxy with public IP --> Amazon Redis (Elasticache)

Notice that there is another good reason to do that (at that time)

As we use node.js client, which don't support Amazon DNS fail over, the client driver don't support dns look up again. If the redis fail, the client driver will keep connect to the old master, which is slave after failed over.

By using HAProxy, it solved that problem.

Now using the latest ioredis driver, it support amazon dns failover.

Afternoons

update for node.js, now ioredis support DNS fail over. If you use the DNS hostname, it can be auto fail over without HAProxy. – Afternoons

Answer 7

This is a solid node script that will do all the dirty work for you. Tested and verified it worked.

https://www.npmjs.com/package/uzys-elasticache-tunnel

How to use Usage: uzys-elasticache-tunnel [options] [command]

Commands:

start [filename]  start tunneling with configuration file (default: config.json)
stop              stop tunneling
status            show tunneling status

Options:

-h, --help     output usage information
-V, --version  output the version number

Usage Example

• start - uzys-elasticache-tunnel start ./config.json
• stop - uzys-elasticache-tunnel stop
• status - uzys-elasticache-tunnel status

Haphazard

Answer 8

Its is not possible to directly access the classic-cluster from a VPC instance. The workaround would be configuring NAT on the classic instance.

NAT need to have a simple tcp proxy

YourIP=1.2.3.4
YourPort=80
TargetIP=2.3.4.5
TargetPort=22

iptables -t nat -A PREROUTING --dst $YourIP -p tcp --dport $YourPort -j DNAT \
--to-destination $TargetIP:$TargetPort
iptables -t nat -A POSTROUTING -p tcp --dst $TargetIP --dport $TargetPort -j SNAT \
--to-source $YourIP
iptables -t nat -A OUTPUT --dst $YourIP -p tcp --dport $YourPort -j DNAT \
--to-destination $TargetIP:$TargetPort

Bela

You gave the same answer in below mentioned post too, which have different requirement. How it can work in the given scenario also ?? #38067408 – Hypocotyl

Answer 9

I resolved using this amazon docs it says you ll have to install stunnel in your another ec2 machine.

https://aws.amazon.com/premiumsupport/knowledge-center/elasticache-connect-redis-node/

Swirl

this is not accessible. Could you add details here about stunnel ? – Whimwham