Home > Storage > ObjectScale and ECS > Product Documentation > ECS with F5 Deployment Reference Guide > F5 Device Redundancy
The BIG-IP DNS and LTM can be deployed as a standalone device or as a group of two or more identical devices for redundancy. In the single-device, or standalone scenario, one device handles all application traffic. The obvious downside with this is that if the device fails applications may experience a complete interruption in service. In a single-site, single-LTM deployment all application traffic will fail if the LTM becomes unavailable.
In multiple-site ECS deployments if each site contains a single LTM and a site’s LTM fails, BIG-IP DNS can direct applications to use the LTM at an alternate site that is a member of the same replication group. So as long as an application can access storage at non-local sites within an acceptable level of performance, if a BIG-IP DNS is in use, deploying a single LTM at each site may allow for significant cost savings. That is, so long as an application is tolerant of any increased latency caused by accessing data using a non-local site, purchasing two or more LTM for each site may not be necessary, providing that a BIG-IP DNS or global load balancing mechanism is in place. Understanding the tradeoffs between implementing single or multiple devices, and the related application performance requirements, is important in developing a deployment best suited for your needs. Also, an understanding of the concept of object owner on ECS, the access during outage (ADO) configuration and impact to object accessibility during temporary site outage (TSO) are all critical to consider when planning for multisite multi-access object namespace. Refer to the ECS Architectural and Overview whitepaper here for more details: http://www.emc.com/collateral/white-papers/h14071-ecs- architectural-guide-wp.pdf.
BIG-IP HA key constructs are:
With two or more local devices, a logical BIG-IP device group can allow application traffic to be processed by more than one device. In a HA setup, during service interruption of a BIG-IP device, traffic can be routed to the remaining BIG-IP device(s) which allows applications to experience little if any interruption to service. The HA failover type discussed in this paper is sync-failover. This type of failover allows traffic to be routed to working BIG-IP devices during failure.
There are two redundancy modes available when using the sync-failover type of traffic group. They are:
It is key to understand that a single virtual IP address can be shared by many virtual servers, however, a virtual IP address may only be associated with one traffic group. If for example a pair of LTMs are configured with many virtual servers that all share the same virtual IP address, all virtual servers can only be served by one LTM at a time regardless of the redundancy mode configured.This means in order to distribute application traffic to more than one LTM device, two or more unique virtual IP addresses must be in use between two or more virtual servers.