| Capacity Manager: | | For every new business project, some computing resources are needed and it is the role of our Capacity Manager to gather enough information for formal a capacity plan to be assembled. | | Even the development phase of a project demands extra hardware and software with increased disk backup space - information that must be quantified and resources that must be purchased. |
| Capacility Plan: | | Disk space is a complex evaluation of how many data objects must be held and how big each data object may be. | | Disk storage suffers from the fact that a 70% full hard disk is close to its operational maximum - 30% free space is a typical operational requirement. | | Computer memory is critical to the performance of any applications, but as memory size increases, the longer it takes to load and unload all the data. | | Processing power is a black art - how many users can be sustained using a 3GHz processor? - the Capacity Manager must have in-depth experience and imperical knowledge of processing options. |
| Capacity Costs: | | Processing power is sold by the calendar month - a nominal 750 hours. | | Memory is sold by the GB per calendar month based on what GB limit is requested. | | Network is sold by 10 GB per calendar month based on what 10 GB limit is requested. | | Data Storage is sold by the GB per calendar month based on what GB limit is requested. | | Data transfer is sold by the GB per calendar month based on what GB limit is transfered. |
| Data Centers: | | Application services operate from at least 2 and normally 3 remote data centers. | | In the event that one data is not available for any reason, another data center will be able to take on the application service and deliver a continuation of the business application service. | | Data is continually replicated from one data center to another, but in the event that a one data center should fail, some transactions that were in progress at the time of the failure will not have been completed and will be lost. | | Other data centers will be able to deliver business continuity with data that was complete and correct at a time just before the equipment failure took place. |
| Backup: | | Backup is continual and incremental - its called data replication uing message switching. | | Recovery is to switch to a parallel data center that is already running - some users will not notice that they are now using a different data center. | | Restart is not implemented as each transaction is limited to 30 seconds. Where such a transaction is to fail for any reason, the user will make the same request again as an automatic restart. |
| Environmental Policy: | | Our Capacity and Demand Managers have taken on the role of environmental officers to publish our environmental policy. | | By replacing older servers with much more efficient servers has seen a significant reduction in our carbon footprint. |
| | | Demand Management: | | While our Demand Manager sets a strategic capacity vision, it is the Capacity Manager who refines that strategy into tactical implementation plans for equipment. | | Our Supply Manager is likely to be responsible for the physical procurement of all applicable hardware to run services. | | Hardware capacity tends to be determined with a one year lead time - we do not want to run out of resources and then begin an urgent procurement cycle. |
| Data Replication: | | An encoded message switch facility continually streams data from one data center database to others in other locations. | | The effect is that in the event of a disaster in one location, a data center in another physical location will be able to take over a continue the application service with the minimum of disruption. | | From time to time applications are switched from one data center to another without people noticing the switch. | | Periodically (monthly) copies of data are taken and copied to spinning hard disk archives as an extra layer of protection. | | Upon request, a CD or DVD copy of application data may be created for a customer - it is understood that disk media may not survive five-years before some corruption is detected. |
|