IMS runs in a wide variety of environments, and supports a selection of technologies to simplify management and installation in each of these. Before you deploy an IMS system, we recommend that you review the options available to help you determine how to get the best results, and the lowest TCO.
Before you start
You should consider the requirements on the system, both initially and a few years down the line. Relevant factors include
- Storage: How much data do you have to store?
- Growth: How much new data will you add each year?
- Users: How many people will access the system at a time?
- Branch access: Will IMS be used by other parts of your organisation in different locations?
- Outside access: To what extent will customers or suppliers use and access the system?
- Business continuity: In the event of a system or site-wide failure, how long can you wait to restore service?
Storage & Growth
This is usually the most significant factor in how to host a system. The decision will affect the cost of storage, as well as the cost of subsequent growth and the limits on that. Try to evaluate how much storage you need now, in 3 years and in 5 years.
The more people use IMS at once, the more powerful the system needs to be to support them. IMS is designed to be as efficient as possible at supporting heavy access, but a whole team uploading and editing files all day will require a system able to keep up with that demand. Conversely, even though you may have 10,000+ users, if only a handful are active at a time, then there is no need to engineer an environment capable of handling them all at once.
Branch & Outside access
If most of your users are working in the same building, it may be best to run your IMS system there too. Local users can benefit from features like Active Directory integration and the File server module. Heavy access from external users is better supported by running your IMS site from a facility with commensurate capacity.
The best solution to recovering IMS fast after a failure is the IMS Sync Module. This feature allows a second IMS instance to act as a backup, that can be activated in moments. It also supports superior data protection through point-in-time recoverable snapshots (by default, daily for a month), and provides an alternative to using a separate backup system, that may require additional storage or licences to be provisioned.
In-house virtual machine deployment
Many businesses now have in-house virtualisation, private or hybrid cloud platforms. Deploying IMS within this environment is a compelling option because it leverages existing investment in hardware and technologies, and can be managed in the same way as other systems.
- Minimal capital expenditure to deploy
- Transparently use existing backup and management
- Simple to add extra CPU/memory resources
- Storage can be incrementally added on-demand (See Automatically growing onto LVM storage)
- Local-network performance and features to on-site users
- Hardware updates are transparent
- Unit cost of storage moderately high
- Marginal cost of extra memory higher than dedicated servers
- Network may be a bottleneck for off-site users
For environments where the total storage used by virtualisation is large compared with the expected size of the IMS, and the majority of usage is internal (or the site has good network connectivity), this is a good option. In instances where the IMS storage would represent 30% or more of the total, it may be more cost-effective to use dedicated hardware.
In-house physical server deployment
Dedicated servers remain strong options for some workloads, and do not require the pre-existing infrastructure of virtual platforms.
- Lowest cost-per-GB (from less than 9p/GB/year including power)
- Can economically provision large amounts of memory
- Does not require infrastructure or specialist IT management
- Support highest performance demands
- Need to purchase and renew hardware
- Entry cost is high for small amounts of storage
- Growth options need to be planned in from the start
- Expansion granularity is limited
For on-site environments where storage demand is high, or there is no existing virtualisation, dedicated servers continue to present strong options. Once a server accounts for a whole tray of disks, it is almost invariably more cost effective to deploy as direct-attached storage, and by default provide more CPU and memory capacity than a typical virtual machine.
There is a continuing trend to use the cloud for application hosting. It provides unparallelled flexibility and scalability - but it is not the best fit for all uses.
- No up-front investment required
- Almost unlimited scalability
- Choice of worldwide locations
- Very high cost of storage (more than 75p/GB/year, not including I/O, data transfer, instance etc.)
- Instance running costs high unless reserved upfront
- Upfront expenditure to reserve negates key flexibility benefit
- Additional cost to achieve comparable responsiveness to other environments
- No access to local-only features
The public cloud is excellent for testing, proof-of-concept, and applications with elastic demand; however it is not well suited to storage-driven uses like Digital Asset Management. For always-on, high storage use, the cost is far in excess of other options.
Hosting with Third Light
Customers who need to reach external users, or with limited internal IT can benefit from fully managed hosting with us.
- Flexible hosting options, for all sizes
- Fully managed for the whole application and platform together
- Single point of contact and support
- Excellent network access for UK and Western Europe
- Choice of managed backup/DR/BC solutions
- Secure hosting facilities
- Geographic backup redundancy
- No access to local-only features
- Network geographically remote from users in the USA, Americas, Asia, Pacific regions
Management by us as the software vendor has a number of advantages, as we know our software and systems the best. If you are outside Western Europe, you should consider the responsiveness benefits of hosting nearby, and if you want to use Active Directory integration or the File server module, you will need to run IMS locally.
Hosting with a co-location/dedicated hosting provider
Many companies specialise in hosting services. From a leased off-the-shelf server, to a custom system in a colocation facility, there are a wide variety of options to suit many needs.
- Choice of purchasing or leasing hardware
- Wide choice of locations all over the world
- Specialist security, network and power delivery
- No access to local-only features
- Introduces a third-party into the support and management of your system
For overseas customers, external hosting avoids the IT needs of in-house systems management, and provides geographic locality for best network performance.
There are two main methods for adding storage to IMS. There are a number of factors that influence which you should use.
The IMS Manager
The IMS manager is a part of the product, and allows flexible, technology-agnostic, hierarchical storage management.
- Supports selection of storage based on usage type
- GUI to monitor and manage usage
- Supports all types of underlying storage
- Can be added after the initial installation
- Requires system-level configuration to prepare and mount storage
- Management required to prevent any single storage location from filling entirely
The IMS Manager provides flexibility and visibility for a variety of storage types. It transparently supports any type of storage that can be mounted in the underlying O/S, be it local, NFS, iSCSI, etc.. It also allows users to gain the performance benefits of solid state disk (SSD) storage without the cost of a pure-SSD installation, by making it possible to use a small, fast datastore for thumbnails, and a larger, slower one for original files.
Automatic Storage Expansion
Automatic storage expansion is achieved by configuring the IMS system at the point of install such that storage additions are utilised without manual system administration work.
See Automatically growing onto LVM storage for details of how to use this option.
- Self-growing, single pool of storage
- Need only provision enough total storage
- Can add very small, granular, increments
- The storage administrator is the only member of the IT team needed to add capacity
- Does not support NFS, or other non-SCSI/ATA-like systems
- Does not support policy-based mapping of usage to disk
This method is ideally suited to VM or SAN-based storage, as these allow allocated capacity to be increased on demand. It also supports using extra virtual disks attached to the VM, or LUNs to the same target transparently, to allow growth past barriers such as the 2TB VMware VMDK limit. Once implemented, the system only needs a reboot to start to use extra capacity.