Cycle 1: Single-region cloudbursting
- Enterprise customer orders single-region “silver” product
a. Customer chooses product (application, SLA and pricing combinations) with “silver” SLA and enters details, such as application, application performance level, pre-authorised scale-up amount, and DC app server and database server ingress IP addresses
- Build contract
a. Design is passed to the ACs and the DM to orchestrate the order
- Enable service
a. Application with cloudbursting service enabled and customer informed
- Customer experiences high traffic on his DC app servers, for example
a. Before SLA thresholds are breached, for example, worsening transaction response times on Enterprise app servers, new app servers are spun up on the BT cloud or a regional partner cloud
b. Overflow traffic is sent to the new app servers, transaction response times ease
c. Further traffic causes the number of new app servers to reach the maximum limit set at order time. All available app servers in danger of overloading “silver” SLA threshold, for example, worsening transaction response times on all available app servers
d. Customer informed of impending overload and asked if he wants to upgrade to “gold” SLA, for example. In this case, “gold” means, add as many app servers as needed to maintain agreed transaction response time.
e. He declines, and any further traffic is queued or declined.
f. As traffic subsides, the overflow app servers are gracefully shut down
g. Graphical dashboards used to help explain the behaviour of the demonstration
- Customer once again experiences high traffic on his DC, for example
a-d. As 4a-d
e. He accepts and new policies are applied. BT partner clouds may be called upon to meet this increase in demand, which may incur further costs of which the customer is kept informed of. No custom is lost, happy customer.
- Graphical dashboards used to help explain the behaviour of the demonstration
Cycle 2 – Agile product and service assembly
- New supported application becomes available
- Technical service design
a. Service designer models new service based on application and “gold” and “silver” SLAs
- Product design
a. Product manager wraps new service into a product bundle with commercial details, such as pricing plans
- Customers able to order the new product
a. This phase is expected to be modelled only. Also, as this cycle is modelled only, we could enrich the service and product portfolios (e.g. by adding security services, other applications and more products) for the catalyst Cycle 3: Multi-regional cloudbursting
This cycle covers the differences from the single-region cycle. In this cycle, SLAs are aware of end-customer location. Here we assume, that the Enterprise customer orders or upgrades to a multi-region cloudbursting product, which permits multi-regional cloudbursting.
This cycle could be merged with cycle 2 showing that instead of a new supported application becoming available, a new application feature (i.e. multi-regional cloudbursting) becomes available for assembly into a new product.
- Enterprise customer orders multi-regional “gold” product
- Build contract
- Enable service
- The customer runs a marketing campaign in Brazil
- Customer experiences high traffic from Brazil on his DC app servers, for example
a. Before SLA thresholds are breached for Brazilian end-customers, for example, worsening transaction response times on Enterprise app servers, new app servers are spun up. The location of these new app servers is decided on SLA and cost. In this case, we assume cost is the same, BT does not have a suitable footprint nearby, and that the regional partner cloud has a significant transaction response time advantage over the BT cloud in the UK
b. Overflow traffic is sent to the new app servers, transaction response times ease
c. As traffic subsides, the overflow app servers are gracefully shut down
- Graphical dashboards used to help explain the behaviour of the demonstration
|