Information Technology Reference
In-Depth Information
LAN/Internet
Researcher
A
Researcher
B
e-Clouds Servers
e-Clouds Te am
e-Clouds
Web Portal
e-Clouds Resource
Manager
e-Clouds
Database
Communication Services (Queues)
Pre-scheduling
Scheduling
Computing Services
Virtual Cluster
Researcher A
Virtual Cluster
Researcher B
CA
CA
CA
CA
CA
CA
CA
CA
Cloud Blod Storage System
e-Clouds
Repository
App
Repository
Data
Repository
e-Clouds
Agent (AG)
App1
User A
App 2
User B
Base
Virtual
Machine
App N
User N
Public laaS Infrastructure
FIGURE 6.1
e-Clouds general architecture.
of executing and monitoring the jobs assigned to its particular machine
inside the cluster.
6.4.2 e-Clouds Back End
6.4.2.1 Data Management
A key to achieving high performance in scientific workload executions is
efficient management of data. For e-Clouds' particular case, it is possible
to differentiate between four types of data: user files, transactional data,
internode messages, and local data. The first two data types are required
to be somehow persistent over time. The other two are associated with a
Search WWH ::




Custom Search