Dell DX6004S DX Object Storage Release Notes - Page 4

Version 5.0, 1.1. Introduction, 1.2. Changes in Release 5.0

Page 4 highlights

Chapter 1. Version 5.0 1.1. Introduction DX Storage is a hardware agnostic, high performance, massively scalable software product that is different from anything on the market today. These release notes contain information about the new features in DX Storage 5.0 as well as important information about using and managing DX Storage. 1.2. Changes in Release 5.0 Release 5.0 has the following features and updates: • DX Storage extends its high performance, massive scalability and data integrity features for public and private clouds to a new type of objects, named objects. Designed for applications that require symbolic names for files or objects, named objects work alongside and are completely compatible with traditional DX Storage objects that are identified by a Universally Unique Identifier (UUID), which are referred to as unnamed objects. For example, you can access named objects stored in a DX Storage cluster using a URL like http://cluster.example.com/marketing/ads/about-object-naming.m4v. • This release adds support for multi-tenancy, which enables enterprises to host multiple companies or departments within a public or enterprise-wide cloud. To define a new tenant, simply add a domain container in the cluster, which gives each tenant an independent name space and authorization lists. Each tenant's domain organizes objects into multiple bucket containers that are each protected with their own security realms and authorization lists. Individual objects can also have attached authorization lists that provide full control over who has permission to view and change the object. • All SCSP operations can be authenticated on a per-object basis using a new metadata header, Castor-Authorization. For example, you can allow users in a security realm named administrators to perform PUT, COPY, APPEND, or DELETE on an object but only users in a security realm named operators can perform a GET or HEAD on an object. For more information, see the DX Object Storage Application Guide. • DX Storage is optimized for SCSP operations on individual objects. If SCSP operations are performed on domains or buckets, which are centralized resources, you should not make bucket POST or DELETE calls on the high availability code path of your client application. Instead, Dell recommends you perform POST and DELETE on buckets using a separate initialization or setup routine that runs less frequently. • A new custom metadata header is supported, x-*-meta-*. See the DX Object Storage Application Guide for more information. • New content caching technology accelerates read access by configuring additional RAM as cache. Frequently read objects automatically replicate to nodes closer to the geographic location from which they are requested. This greatly reduces latency and improves throughput. DX Storage automatically identifies high-demand objects and dynamically caches those objects. When the demand wanes, the objects are automatically removed from cache, leaving room for higher-demand objects that are then moved into cache. Copyright © 2010 Caringo, Inc. All rights reserved 1 Version 5.0 December 2010

  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8

Copyright © 2010 Caringo, Inc.
All rights reserved
1
Version 5.0
December 2010
Chapter 1. Version 5.0
1.1. Introduction
DX Storage is a hardware agnostic, high performance, massively scalable software product that is
different from anything on the market today. These release notes contain information about the new
features in DX Storage 5.0 as well as important information about using and managing DX Storage.
1.2. Changes in Release 5.0
Release 5.0 has the following features and updates:
DX Storage extends its high performance, massive scalability and data integrity features for public
and private clouds to a new type of objects,
named objects
. Designed for applications that require
symbolic names for files or objects, named objects work alongside and are completely compatible
with traditional DX Storage objects that are identified by a Universally Unique Identifier (UUID),
which are referred to as
unnamed objects
.
For example, you can access named objects stored in a DX Storage cluster using a URL like
.
This release adds support for
multi-tenancy
, which enables enterprises to host multiple
companies or departments within a public or enterprise-wide cloud. To define a new tenant,
simply add a
domain
container in the cluster, which gives each tenant an independent name
space and authorization lists.
Each tenant's domain organizes objects into multiple
bucket
containers that are each protected
with their own security realms and authorization lists. Individual objects can also have attached
authorization lists that provide full control over who has permission to view and change the object.
All SCSP operations can be authenticated on a per-object basis using a new metadata header,
Castor-Authorization
. For example, you can allow users in a security realm named
administrators
to perform PUT, COPY, APPEND, or DELETE on an object but only users in a
security realm named
operators
can perform a GET or HEAD on an object.
For more information, see the DX Object Storage Application Guide.
DX Storage is optimized for SCSP operations on individual objects. If SCSP operations are
performed on domains or buckets, which are centralized resources, you should not make bucket
POST or DELETE calls on the high availability code path of your client application.
Instead, Dell recommends you perform POST and DELETE on buckets using a separate
initialization or setup routine that runs less frequently.
A new custom metadata header is supported,
x-*-meta-*
. See the DX Object Storage
Application Guide for more information.
New content caching technology accelerates read access by configuring additional RAM as
cache. Frequently read objects automatically replicate to nodes closer to the geographic location
from which they are requested. This greatly reduces latency and improves throughput.
DX Storage automatically identifies high-demand objects and dynamically caches those objects.
When the demand wanes, the objects are automatically removed from cache, leaving room for
higher-demand objects that are then moved into cache.