Dell POWEREDGE R710 - Manual (Page 7)

Dell HPC Tiered Storage Solution
Page 5
Introduction
Clusters have become one of the most popular architectures for High Performance Computing (HPC)
today.
(1) Along with the increasing popularity in HPC architectures, the HPC storage market has
witnessed an explosion in data growth, presenting an increasing concern for data management.
Enterprise IT departments are searching for solutions to address the data growth and
regulatory/compliance requirements, while limiting the effect on IT budgets and resources. Data
archiving and management are key emergent areas to address the rising data storage needs effectively.
Customers are seeking effective solutions to:
 Organize data storage and retrieval into separate tiers for cost management and storage space
efficiency.
 Simplify data storage processes.
 Focus on shrinking expensive primary storage requirements and migrating data to cheaper
secondary storage tiers.
HPC Customers typically have three kinds of needs for an HPC storage solution.
Scratch Space (tier1): high throughput and scalable cluster working space. This solution is provided by
the Dell | Terascala HPC Storage Solution (DT-HSS), which offers a high throughput scale-out storage
appliance based on the Lustre file system and Dell PowerVault storage arrays.
(2)
Primary Storage (tier2): reliable, cost effective and good performance storage for user data. This
solution is provided by the Dell NFS Storage Solution (NSS) that uses the NFS file system on top of the
Red Hat Scalable File System (XFS) with Dell PowerVault storage. It provides an easy to manage,
reliable, and cost-effective solution for unstructured data.
(3)
Long-Term Storage (tier3): low cost, high capacity storage for long-term retention. Long-term storage
is disk based, tape based, or a combination of both.
The Dell Tiered Storage Solution (TSS) enables Dell customers to manage the archival and movement of
their data between these tiers.
More specifically, the TSS leverages sophisticated, policy-based data management automation
processes to move data from tier to tier, leaving behind a small stub file containing the file metadata
and moving the data blocks to a lower tier (commonly referred to as HSM). Additionally, you can use it
for archiving purposes where not file stub is left behind and a new method of accessing the data is
required. Typically, frequently used data can be moved to a faster tier, while infrequently used data
can be moved to slower media at lower tiers. Correspondingly, as files are used more often, they can
reside on the first or fastest tier and then moved down to the slowest or lowest tier as they fall out of
use, but must be retained to meet regulatory or compliance requirements.
The following sections describe the TSS architecture in detail with focus on the CommVault Simpana 9
Data Management Software
(4) in a Hierarchical Storage Management (HSM) implementation.
Subsequent sections discuss a selection of typical use cases scenarios of an HSM solution focusing on
performance and best practices of implementation for the solution. An extensive appendix covers
detailed steps on configuring a TSS.