AWS EBS vs EFS for data-directory

Hello everyone.
I am going to set my data directory on aws ebs but I am afraid that if my current Availability zone goes down, ebs will also down and if I start another ec2 I wouldn’t be able to attach the existing ebs to my new ec2 to workaround my previous data. So How to achieve high availability in such case.
2nd possible solution is to work with EFS because even if an AZ is down I can attach efs to other ec2 in another AZ. but according to documentation EFS is not recommended FOR Embedded datA AND BLOBSTORE. why EFS is not recommended can you plz explain.?
Will you sugesst to use EFS in my case.
Thank you

What our documentation calls Embedded Data (aka data-directory) is data that must not be shared between nodes, hence there’s no need to have it available between different AZs. You’re fine with using EBS, it’s just that in case of one AZ going down, all that embedded might be lost. In case of some data, like Elasticsearch indices, they will be recreated on new instance boot. For other data, like logs, make sure you collect them for troubleshooting (in CloudWatch or any other services of your choosing). However, if you are using the embedded H2 database make sure you make regular backups.
EFS is not recommended because of it’s performance issues, that are also documented by Elasticsearch. Whether or not EFS will cause any issues to you depends on how busy is your Nexus Repository instance. Unfortunately I cannot put any numbers to this statement.


Our help site mentions it (System Requirements) and explains why.