Home > Storage > ObjectScale and ECS > Product Documentation > ECS with Kemp LoadMaster Deployment Reference Guide > Kemp LoadMaster deployment options
There are many deployment options based on the environment and customer requirements. This section will cover some of the more common deployments and configuration options.
Deployments using a single LoadMaster delivers the necessary functionality to load balance Dell ECS but does introduce a single point of failure into the environment. This configuration although supported is not recommended due to the possibility of a system outage should this single unit going offline for either planned or unplanned maintenance.
The recommended deployment for a single site is running the LoadMasters in an active/passive HA configuration. HA enables two physical or virtual machines to become one logical device. Only one of these units is active and handling traffic at any one time while the other unit is a hot standby (passive). This provides redundancy and resiliency, meaning if one LoadMaster goes down for any reason, the hot standby becomes active, therefore avoiding any downtime.
There are some prerequisites to be aware of before setting up HA:
When Dell ECS is deployed across multiple locations and there is a requirement for site reliance, Kemp GEO can be leveraged to provide this availability. GEO offers the ability to move past the single data center, allowing for multi data center High Availability (HA). Even when a primary site is down, traffic is diverted to the disaster recovery site. Also in GEO is the ability to ensure clients connect to their fastest performing and geographically closest data center.
GEO can be deployed in a distributed (active/active) high availability configuration, with multiple GEO LoadMasters securely synchronizing information. Introducing GEO into existing Authoritative Domain Name Services (DNS) requires minimal integration work and risk, allowing you to fully leverage the existing DNS investment.
There is generally no special configuration required to support load balancing strategies within ECS. ECS is not aware of any Kemp LoadMaster systems and is strictly concerned, and configured with, ECS node IP addresses, not virtual addresses of any kind.
Regardless of whether the data flow includes a traffic manager, each application that uses ECS will generally have access to one or more buckets within a namespace. Each bucket belongs to a replication group. It is the replication group that determines both the local and potentially global protection domain of its data, as well as its accessibility. Local protection involves mirroring and erasure coding data inside disks, nodes, and racks that are contained in an ECS storage pool. Geo-protection is available in replication groups that are configured within two or more federated VDCs. They extend protection domains to include redundancy at the site level.
Buckets are generally configured for a single object API. A bucket can be an S3 bucket, an Atmos bucket, or a Swift bucket, and each bucket is accessed using the appropriate object API. As of ECS version 3.2 objects can be accessed using S3 and/or Swift in the same bucket. Buckets can also be file enabled. Enabling a bucket for file access provides additional bucket configuration and allows application access to objects using NFS and/or HDFS.
Application workflow planning with ECS is generally broken down to the bucket level. The ports associated with each object access method, along with the node IP addresses for each member of the bucket’s local and remote ECS storage pools, are the target for client application traffic. This information is what is required during Kemp LoadMaster configuration. In ECS, data access is available using any node in any site that serves the bucket. In directing the application traffic to a Kemp LoadMaster virtual service, instead of directly to an ECS node, load balancing decisions can be made which support HA and provide the potential for improved utility and performance of the ECS cluster.