The ETERNUS DX S4 models DX100, DX200, DX 500 & DX600 all support both file and block storage provisioning.
Data is currently growing at a yearly average rate of 50%. Some industries report even up to 100%. Unstructured data, i.e. files, are a major factor in these percentages. In contrast to structured data in databases and OLTP (Online Transaction Processing) applications, it is much more difficult to anticipate the necessary storage capacity for files as their size can vary greatly.
File and block storage have been traditionally separated on dedicated systems. File storage, utilising NAS (Network Attached Storage) architecture, has been preferred in many cases due to its deployment simplicity. On the other hand, block level storage systems with SAN (Storage Area Network) architecture, provide unmatched flexibility.
They can fully concentrate on data handling whereas higher level data management, such as file services, runs outside, typically on servers.
In recent years, the boundaries between both types have started to diminish and now combine the benefits of both. Unified storage systems, which include file access on top of block access, provide higher consolidation potential by hosting structured data from databases plus unstructured data in the form of files.
The main benefits of such a unified approach are:
■ Less operational and training expenditure
■ Use of the same software features for both, block and file operation
■ Higher storage resource consolidation
Fujitsu ETERNUS DX S4 is offering unified storage systems for the scalable entry and mid-range segments. Customers can choose between SAN-only and unified systems. Moreover, SAN-only models can be easily upgraded to the unified models.
When deploying the ETERNUS DX S4 systems for unified storage, the architecture is slightly adapted in order to handle both SAN and NAS functionalities.
In addition to the conventional SAN control, the same controller module can also process the NAS, as shown in the next figure.
Distributed file system is used to control the file management capability in the ETERNUS DX S4. Network File System (NFS) and Common Internet File System(CIFS) are supported as file sharing protocols. In the third version of S4 release more protocols will be supported, such as HTTP, HTTPS, FTP, SCP, SFTP and WebDAV.
The file system is installed on each controller module and performs shared control for a single NAS volume (Shared VOL). Hence, activeactive connections can be implemented as cluster systems in CM#0/CM#1 from CIFS/NFS client.
The main features of NAS volumes are as follows:
■ Thin Provisioning Volumes (TPVs) are used as NAS volumes
■ Formatting a single TPV is performed with a single file system
■ The Thin Provisioning Pool (TPP), to which the TPV belongs, can be selected
■ SAN and NAS volumes can be mixed in the same TPV ensuring efficient operation
ETERNUS DX S4 supports three different topologies for the connection to NAS volumes:
|Single Path Connection This topology has the client access the NAS volume via a single path. Each NAS volume is connected via only one controller module. In the event of a CM failure, failover is not possible and the system is down.In order to provide higher availability and reliability, the following multi-path connection topologies are also supported:|
|Active-Active Multi-Path Connection In this case the NAS volume can be accessed by the client via two parallel paths using both controller modules. If one CM fails, information is taken over by the other CM and the failover operation is carried out. After maintenance for the failed hardware has been completed, the connection topology can return to the previous state.This topology allows load balancing between both CMs.|
|Active-Standby Multi-Path Connection In this topology the client can access the NAS volume only via the active path (CM#0 in the figure below). The other path (CM#1) is in standby mode. If the active connection fails, the client detects the failure and the other CM takes over. Failover operation is then carried out. After maintenance for the failed hardware has been completed, the connection topology can return to the previous state. This topology prevents performance degradation, even when one CM is not available.|
As the file management consumes additional system resources, the unified storage systems have to provide higher performance in order to avoid quality degradation. The total system cache is thus increased for unified models and dedicated cache is assigned for the file management.
The cache area dedicated for SAN control ensures data writing on a moment-to-moment basis regardless of control from the application. On the other hand, the cache area for NAS file system writes data only when a Sync request is received from the client, according to RFC3530.
In the same way as SAN control, data integrity function (DIF) control is performed for NAS I/O to improve data reliability. Data copying as well as DIF generation, removal, and checking are carried out between operating systems using hardware mechanisms without creating any additional CPU load.
If a DIF error occurs, the information is sent to the access host. For READ I/O, re-reading from the disk is performed internally.
Quota limits are set for the number of files that can be created and the total file data size for each file owner and ownership group. The client is notified when the specified value is exceeded.
This function will be supported in the next version of the S4 release.
In addition to traditional data management, a unified storage system also performs file management functions, which requires enhanced security functions.
In order to ensure that only authorized users can access the system, the following authentication methods are implemented depending on the protocol used.
When CIFS protocol is used to access the device and user management is performed for directories and files, authentication by an ActiveDirectory authentication server (Kerberos authentication method) must be performed.
If NFS protocol is used to access the device and user management for directories and files, authentication by an LDAP server (such as an OpenLDAP server) is required.
This function allows data restoration should a data volume be lost. It collects and saves a copy of the entire directory in a specified volume. The stored data can then be used in the case of data loss. There are two options for this functionality:
Volumes used as NAS are cloned and copy source volumes are then restored. The whole operation is done independently of the operating systems and applications without affecting the server.
Prompt data recovery of each file or directory is provided if a file or directory is deleted or updated by mistake. This function can be configured for individual users or for each
application. This function will be supported in the next version of the S4 release.
NAS Data Protection functions require the use of ETERNUS SF AdvancedCopy Manager (ACM) software.
Link aggregation is supported in order to improve fault tolerance and increase overall performance. It combines multiple ports and aggregates them together in order to provide link resiliency or higher throughput. Load balancing is supported as well.