Efficient content delivery to end users is the basic value proposition for investing in content caches. Caches essentially extend the Web by bringing content closer to end users.
However, simply installing multiple cache devices doesn't guarantee a good return on investment. For this to happen, the cache devices must be deployed and managed as part of a total content networking solution that provides for the following:
Content Distribution: The managed, intelligent distribution of content from origin servers to strategically placed network caches
Device Configuration Management: The ability to easily manage widely distributed cache devices from a central location
Request Routing: The directing of user requests to the optimal caches capable of filling the requests
Content Delivery: The delivery of cached data, including static, dynamic, and streaming content, to end users
Billing and Accounting: When required, the ability to track and charge for bandwidth usage, object hits, etc.
All of these elements are required by a true content distribution and delivery network.
The following example illustrates why only installing multiple cache devices doesn't create an effective content distribution solution.
Example: A multinational corporation installs multiple cache devices in a configuration similar to that shown in Figure 1 with plans to provide business-related Web and company training content to its employees.
Unfortunately, the administrative burden of maintaining the content network is so great that its potential value is substantially reduced. The following graphic summarizes the problems:
Figure 2
For an organization to realize a good return on investment, its content network solution must address these problems.
The following sections outline the basic requirements of an effective content network solution.
A cost-effective content network solution must let content administrators control what content is cached, when it is cached, where it is cached, and how long it is retained in cache.
For specific information on Velocity management suite's content management features, see Content Controller Overview .
Content administrators must be able to specify what is cached on their content networks without having to configure each cache individually. This is the only way they can ensure that caching resources are used effectively and that content is available when needed.
To be successful, content administrators must be proactive. They must track content usage and detect trends. This lets them predict and project content usage demands so they can pre-position high-demand content to major population areas and capture content business when the demand hits without incurring the cost of using premium bandwidth.
Enterprise content administrators must be aware of their organization's business-critical content, such as training material and important announcements. They can then prepopulate caches in key locations so employees don't waste their time waiting for training to be available.
An effective content network solution must let content administrators automatically
Most cache devices rely on request-driven caching---objects are cached only when users specifically request them.
If the first request for a large object must be filled from the Web, there could be a substantial delay while the object is retrieved. Additionally, if the object is requested during peak traffic hours, the cost of retrieving it could be many times higher than if it were retrieved and cached while rates are low.
Therefore, if content administrators know that specific data is going to be in demand, they will want to get it cached on the edge before it is needed and while rates are low.
An effective content network solution must let content administrators schedule the prepopulation of caches during the hours when rates are lowest.
Request-driven caching has other drawbacks.
For example, a company might want to reserve the resources of high-end cache devices for only delivering training videos at various company locations where the devices are deployed.
An effective content network solution must let content administrators specify exactly where which objects will be cached.
Most caching devices retain content in cache based on frequency of usage and other system-specific configuration settings, such as available disk space and time-to-live (TTL) object header information. When these arbitrary threshholds are reached, objects are automatically removed from cache.
Such arbitrary recycling of cached content often removes objects that aren't actually outdated. Also, it doesn't allow content administrators to retain objects that they know will be needed again soon and/or might be relatively expensive to retrieve again the next time someone needs them.
An effective content network solution must let content administrators specify how long objects are retained in cache, how often they are verified for freshness, and when they are removed.
As discussed earlier, simply installing multiple cache devices doesn't guarantee a good return on investment. In fact, as the number of devices increases, it becomes increasingly difficult to manage the network as a cohesive unit.
While some management tasks, such as IP address changes apply to only single cache devices, other tasks, such as enabling or disabling caching services, might apply to multiple cache devices.
An effective content network solution must let administrators
For more information on Velocity management suite's device management features, see System Controller Overview .
Although request routing isn't required by the Velocity solution, it can be an important part of your content network.
The term request routing describes various routing technologies that direct end user requests to the optimum cache. There are three technologies that can be used for this purpose. These are described in order of effectiveness in the sections that follow.
An effective content network solution that can locate specific cached content will interface with each of these routing technologies.
The key component in application layer routing is a device such as a layer 4-7 switch that transparently routes traffic based on
Application-layer routing lets administrators create content domains---sets of caches that serve specific content. The result is reduced management and equipment costs.
Transport layer routing uses only the information available in layer 4 of the OSI Reference Model, which includes the client's IP address and port number.
Layer 4 information can be used in conjunction with various policies and metrics to direct requests to specific caches.
However, depending on the metrics and algorithms employed, the actual method for determining the correct cache can be quite complex. This results in latency delays for the first content request served using this method.
Dynamic DNS routing is the easiest and simplest to use. However, it enables only domain-level redirection, as contrasted with the object-level redirection provided by the other two methods.
This method routes all requests for a domain to the same cache. For example, all content from www.myuniversity.edu would be requested from the same cache, even though requests to www.myuniversity.edu/curriculum might be better served from a different cache.
Additionally, because DNS entries have very short time-to-live values, many service providers utilize DNS caching to reduce DNS server load. This often results in stale routes and poor performance.
In addition to managing the caching infrastructure and the objects that are cached, a complete content network solution must let administrators gather system and content statistics, including information on how individual devices are performing, what content is most used, where it is being delivered, and so on. They can then leverage this information for effective content management and other business-driven purposes, such as trending, cost tracking, and billing for content usage.
For more information on Velocity management suite's statistical reporting features, see Content Accountant Overview .