Ceph

Since 2016 we have been providing Cluster Ceph assistance, design and deployment.

We learn from and we are always in contact with the most important European representatives for the development, implementation and maintenance of Ceph clusters.

Ceph is a registered trademark of Red Hat (now IBM).

To understand what Ceph is and its importance, let’s compare it with traditional storage.

 

Assistenza Ceph

A COMPARISON WITH TRADITIONAL STORAGE

Traditional storage systems are based on the Client – Server concept and are no longer efficient enough to meet today’s performance, scalability and reliability demands.

Not to mention the persistence of the data.

1

A single ip to talk with, often resolved using virtual ip or multipath protocols

2

A client who does not know the positioning of the data

3

All I/O processes go through a single machine

4

Physical limitations on the number of

  • Discs
  • Network
  • Controller
  • CPU
  • Memory

5

Protocols are not adequate for the large distributed systems that we need to manage today

A COMPARISON WITH TRADITIONAL STORAGE

Traditional storage systems are based on the Client – Server concept and are no longer efficient enough to meet today’s performance, scalability and reliability demands.

Not to mention the persistence of the data.

Traditional storage includes:

    1. A single ip to talk with, often resolved using virtual ip or multipath protocols

 

    1. A client who does not know the positioning of the data

 

    1. All I/O processes go through a single machine

 

    1. Physical limitations on the number of

 

  • Discs
  • Network
  • Controller
  • CPU
  • Memory

 

  1. Protocols are not adequate for the large distributed systems that we need to manage today

Market needs

Currently the market needs for storage require much more, given the application area devoted to the use of the Cloud, Virtual Machines and Containers:

    1. File based storage (POSIX) (container)

 

    1. Block storage (Virtual Machines)

 

    1. Object storage (Applications / Unstructured data / Cloud)

 

And to store this data we need

    1. No limits to the increasing of storage

 

    1. A mix of usable hardware, no vendor lock-in and no proprietary protocols

 

    1. Low operating cost

 

    1. Low maintenance cost

 

Talking about operating and maintenance costs, we want to automate storage as much as possible with

    • Automatic data migration

 

    • Automatic data redistribution

 

    • Scalability

 

    • Data guarantee by default

 

So the new concept of Storage (and therefore Ceph) provides a new design

    • Scalable

 

    • With no Spof (Single Point of Failure)

 

    • Software based / Hardware agnostic

 

    • Flexible

 

    • Unified

 

Safe against faults by design

ceph's unified storage

ITServicenet offers consultancy, design and maintenance of distributed storage systems based on Ceph.