26 April 2008 - 11:48pm, Arkady Godin (not verified) said:
Hi Michael,
While applauding a direction Appistry EAF has taken by attacking the core issue of the applications, which is a service virtualization, I wonder whether there is a need to scale "across services" per a single application and "across services" "across apps".
This brings a need to look at a layer below the services, which is the workload. Applications frequently have a need to process different sizes of workloads, corresponding to different data sources, at about the same time. This requires treating data as logical partitions.
Additional complexity is a need to support prioritization of various data sources in a Pub/Sub sense as subscribers might need to get a product, containing more data and requiring more processing, earlier than a simpler product to produce. This might require utilization of a BPM handling prioritization of scheduling the execution of Virtualized Services.
Does it all mean that there is a need to have a Workload Virtualization Manager?
Is Appistry EAF moving in this direction?
Workload Virtualization. Is it relevant for Service Virtualizati
Hi Michael,
While applauding a direction Appistry EAF has taken by attacking the core issue of the applications, which is a service virtualization, I wonder whether there is a need to scale "across services" per a single application and "across services" "across apps".
This brings a need to look at a layer below the services, which is the workload. Applications frequently have a need to process different sizes of workloads, corresponding to different data sources, at about the same time. This requires treating data as logical partitions.
Provided that algorithms for processing various data sources differ, Event Processing Fabric ("Distributed CEP Engines over the Fabric" - term introduced by Tim Bass in http://thecepblog.com/2007/11/16/clustered-databases-versus-virtualizati...) needs to start providing Workload Virtualization.
Additional complexity is a need to support prioritization of various data sources in a Pub/Sub sense as subscribers might need to get a product, containing more data and requiring more processing, earlier than a simpler product to produce. This might require utilization of a BPM handling prioritization of scheduling the execution of Virtualized Services.
Does it all mean that there is a need to have a Workload Virtualization Manager?
Is Appistry EAF moving in this direction?
Respectfully,
Arkady Godin