公开(公告)号：GB2503463A

公开(公告)日：2014-01-01

申请号：GB201211388

申请日：2012-06-27

Applicant: IBM

Inventor： SPATZIER THOMAS ,  ARWE JOHN E ,  SCHWERTLE ISABELL ,  MOSER SIMON DANIEL ,  LINDQUIST DAVID B ,  DROSS FELIX ,  LE HOANG ANH ,  KOKHLIKYAN NARINE ,  CHODOROWSKI MAREK ,  BREITER GERD

IPC: G06F9/44 ,  G06F9/50

Abstract: A service, requested by a server system 205, 206, is defined as nodes in a service model 211 in a management engine system 201. Each node defines the type of resource required for that node as an abstract method. The management engine 209 find resource manager systems 202, 203, which implement the resource, by searching a service provider catalogue 210. The management engine requests the resource manager to provide its application programming interface (API). The management engine overrides the abstract interface in the node with the interface of the selected resource manager to provide the specific resource. The resources may be data storage, application programs, computing resources, networking components or middleware.

公开(公告)号：GB2503464A

公开(公告)日：2014-01-01

申请号：GB201211389

申请日：2012-06-27

Applicant: IBM

Inventor： SPATZIER THOMAS ,  SCHWERTLE ISABELL ,  MOSER SIMON DANIEL ,  BREITER GERD

IPC: G06F9/50

Abstract: A server 205, 206 makes a request for a service in a distributed computing system 200, such as a cloud computing environment. A management engine system, running a management engine 209, 201, analyses a model 211 of the requested service. The service model contains a graph of nodes. Each node has a resource type and may have management rules. For each node, the management engine system uses a resource instance catalogue 250 and a service provider catalogue 210 to find a resource manager 214, 219, 222, which can provide a resource instance, which matches the type and rules of then of the node. The resource manager then creates the resource instance and allocates it to providing the requested service. The rules may include load requirements for resource instances, which may be updated based on the actual monitored loads of the instances.