Choose between redundancy at the application level or at the virtual machine level

Redundancy at the application level

In this type of solution, only application data are replicated. And only the application is restared in case of failure.

With this solution, restart scripts must be written to restart the application.

We deliver application modules to implement redundancy at the application level. They are preconfigured for well known applications and databases. You can customize them with your own services, data to replicate, application checkers. And you can combine application modules to build advanced multi-level architectures.

This solution is platform agnostic and works with applications inside physical machines, virtual machines, in the Cloud. Any hypervisor is supported (VMware, Hyper-V...).

• Solution for a new application (restart scripts to write): Windows, Linux

Redundancy at the virtual machine level

In this type of solution, the full Virtual Machine (VM) is replicated (Application + OS). And the full VM is restarted in case of failure.

The advantage is that there is no restart scripts to write per application and no virtual IP address to define. If you do not know how the application works, this is the best solution.

This solution works with Windows/Hyper-V and Linux/KVM but not with VMware. This is an active/active solution with several virtual machines replicated and restarted between two nodes.

• Solution for a new application (no restart script to write): Windows/Hyper-V, Linux/KVM

More comparison between VM HA vs Application HA

How the SafeKit mirror cluster works?

Step 1. Real-time replication

Server 1 (PRIM) runs the application. Clients are connected to a virtual IP address. SafeKit replicates in real time modifications made inside files through the network. 

The replication is synchronous with no data loss on failure contrary to asynchronous replication.

You just have to configure the names of directories to replicate in SafeKit. There are no pre-requisites on disk organization. Directories may be located in the system disk.

Step 2. Automatic failover

When Server 1 fails, Server 2 takes over. SafeKit switches the virtual IP address and restarts the application automatically on Server 2.

The application finds the files replicated by SafeKit uptodate on Server 2. The application continues to run on Server 2 by locally modifying its files that are no longer replicated to Server 1.

The failover time is equal to the fault-detection time (30 seconds by default) plus the application start-up time.

Step 3. Automatic failback

Failback involves restarting Server 1 after fixing the problem that caused it to fail.

SafeKit automatically resynchronizes the files, updating only the files modified on Server 2 while Server 1 was halted.

Failback takes place without disturbing the application, which can continue running on Server 2.

Step 4. Back to normal

After reintegration, the files are once again in mirror mode, as in step 1. The system is back in high-availability mode, with the application running on Server 2 and SafeKit replicating file updates to Server 1.

If the administrator wishes the application to run on Server 1, he/she can execute a "swap" command either manually at an appropriate time, or automatically through configuration.

More information on power outage and network isolation in a cluster.

Typical usage with SafeKit

Why a replication of a few Tera-bytes?

Resynchronization time after a failure (step 3)

• 1 Gb/s network ≈ 3 Hours for 1 Tera-bytes.
• 10 Gb/s network ≈ 1 Hour for 1 Tera-bytes or less depending on disk write performances.

Alternative

• For a large volume of data, use external shared storage.
• More expensive, more complex.

Why a replication < 1,000,000 files?

• Resynchronization time performance after a failure (step 3).
• Time to check each file between both nodes.

Alternative

• Put the many files to replicate in a virtual hard disk / virtual machine.
• Only the files representing the virtual hard disk / virtual machine will be replicated and resynchronized in this case.

Why a failover ≤ 32 replicated VMs?

• Each VM runs in an independent mirror module.
• Maximum of 32 mirror modules running on the same cluster.

Alternative

• Use an external shared storage and another VM clustering solution.
• More expensive, more complex.

Why a LAN/VLAN network between remote sites?

• Automatic failover of the virtual IP address with 2 nodes in the same subnet.
• Good bandwidth for resynchronization (step 3) and good latency for synchronous replication (typically a round-trip of less than 2ms).

Alternative

• Use a load balancer for the virtual IP address if the 2 nodes are in 2 subnets (supported by SafeKit, especially in the cloud).
• Use backup solutions with asynchronous replication for high latency network.

How the SafeKit farm cluster works?

Virtual IP address in a farm cluster

On the previous figure, the application is running on the 3 servers (3 is an example, it can be 2 or more). Users are connected to a virtual IP address.

The virtual IP address is configured locally on each server in the farm cluster.

The input traffic to the virtual IP address is received by all the servers and split among them by a network filter inside each server's kernel.

SafeKit detects hardware and software failures, reconfigures network filters in the event of a failure, and offers configurable application checkers and recovery scripts.

Load balancing in a network filter

The network load balancing algorithm inside the network filter is based on the identity of the client packets (client IP address, client TCP port). Depending on the identity of the client packet input, only one filter in a server accepts the packet; the other filters in other servers reject it.

Once a packet is accepted by the filter on a server, only the CPU and memory of this server are used by the application that responds to the request of the client. The output messages are sent directly from the application server to the client.

If a server fails, the SafeKit membership protocol reconfigures the filters in the network load balancing cluster to re-balance the traffic on the remaining available servers.

Stateful or stateless applications

With a stateful application, there is session affinity. The same client must be connected to the same server on multiple TCP sessions to retrieve its context on the server. In this case, the SafeKit load balancing rule is configured on the client IP address. Thus, the same client is always connected to the same server on multiple TCP sessions. And different clients are distributed across different servers in the farm.

With a stateless application, there is no session affinity. The same client can be connected to different servers in the farm on multiple TCP sessions. There is no context stored locally on a server from one session to another. In this case, the SafeKit load balancing rule is configured on the TCP client session identity. This configuration is the one which is the best for distributing sessions between servers, but it requires a TCP service without session affinity.