Geek Answers Handbook

Single point of failure when scaling an application through AWS

We have a Rails-based application , the deployment infrastructure is tied to AWS . The current scheme included the following layers:

  • load balancer (HaProxy)
  • Rails Application (EC2) x2
  • MySQLd Database (Master-Slave EC2)
  • Redis source processes, DelayedJob
  • Wowza Multimedia Server (EC2)
  • S3 Storage (General)

There are 3 SPFs: load balancing, database, media server.

My questions are about redundancy, how can I reduce the SPF:

  • load balancer. We have a plan to install a secondary load balancer, but the domain name remains the same. Does DNS A / AAAA roundrobin failover work well in this case? Is the AWS balancer used correctly?
  • MMM (Multi-Master Replication Manager)? Rails (/ )?
  • Wowza media server, - HA ?
+5
3

, , , .

SPF - , Amazon. AWS , , , , , , 100%. , , 1 (, VPS ).

: AWS , // , - , , ?

, , : DNS A/AAAA , , , , SRV/MX. , AWS , DNS, IP. CAN , DNS API, . , DNS , DNS, 50% 100%, 1 IP ( , 2 A), IP-, .

, AWS, , 2 IP- ( ) . , , 0% , 1 IP , 50% .

2 , / . , BOTH / ONE-. , , AWS;)

MMM - , Rails. , .. data​​strong > , , , , , .. MMM , , , . .

- Wowza, . Wowza RTSP (UDP TCP), HAProxy , TCP. Keepalived, , UDP ( IVPS/LVS). , Keepalived , .

, " " AWS- , S3. SPF, , AWS, , Eucalyptus/Cloud.com/Openstack/GlusterFS. , , , : " , X- , Y ".

+5

:

1) : ha_proxy . Amazon ha_proxy. ha_proxy, .

2) : , MySQL, , .

3) Wowza: Wowza ha_proxy , Wowza

+1

Scalarium , SPF, Rails 12.

  • Amazon Elastic Load Balancer ha_proxy. .

  • MySQL . . , . , - .

  • ha_proxy Wowza.

+1

More articles:


All Articles