Data Storage Infrastructure

Data Centers operated by CESNET are based on several technologies. It goes about common Hierarchical Data Storage (HSM), object storage utilizing Ceph and disk arrays utilizing RAID technology, see below. Data Centers are located only within Czech Republic so all your data lies thereto. Distributed model of Data Centers allows us to increase data redundancy due to usage of geographical replicas. That model brings the advantage that your data are resistant to the data media failure, nature disaster or user mistake. Another advatage is that you can use the Data Center which is the closest to your location.

CESNET Data Storages utilizing folowing technologies:

HSM - Hierarchical Storage Management - Hierarchical model consists of two or more layer of data media. CESNET Data Storage utilizes two layers (hard disks and tape library). Frequently used data are then stored in fast media, (SAS “Serial Attached SCSI” HDD). HSM takes care of moving rarely used data to the slower/cheaper tier (Tape Library) from the disk array. HSM technology provides us with large capacity in comparison with disk-only data storage while keeping same investment cost.

Disk arrays - RAID technology - Redundant Array of Independent hard disks – multiple disk array of independent hard disks . That is the method to secure your data against disk media failure. The core of RAID technology is storing your data onto more independent hard disks or disk arrays. In case of failure are corrupted data recalculated/rebuilt with specific algorythms. Our data storage utilizes RAID 6, so we can rebuild your data even if two hard disks from eight have failed.

CEPH - Ceph is the software solution for providing object storage solution. Object storage differs from common file-oriented data storages. Object storage stores data as an objects not like data blocks. Metadata is associated directly to the particular object. This system eliminates hierarchical structure utilized within common data storages. All data are then stored in flat name space which is easy scalable. Ceph is opensource system. It consists of object store, block store and distributed file system. Ceph supports object access (RADOS), exported block device and file access. Data stored within Ceph are replicated to multiple nodes, based on defined replication policy.

Data Center in Ostrava is HSM-like and it has two tiers. Tier 1 are disk arrays consisting of 755pcs of 10 TB SAS HDD. Tier 2 is represented by Tape Library TS4500 with twelve drives served with two robots. Tape Library has 1167pcs of tapes, each has capacity 15 TB. Tier 1 and Tier 2 are operated by GPFS file system. New and actively used data are on the Tier 1 and subsequently LIFO-like (Last In First Out) migrated to the Tier 2. The access to the data is served by three front end servers.

• VO_storage accounts are homed here.
• DU4 has several migration policies tape, tape_tape a disk_only
• Virtual organization has archives and backup drives here.
• DU4 contains also data of ownCloud, Filesender
• Fotos

Jihlava Data Storage is consisting of three disk arrays. Each disk array has 38 RAID sets (RAID6 - 8+2) consisting of 10 TB NL (Near Line) HDD. Total usable capacity is then 9.36 PB. Metadata has own disk array consisting of SSD discs (144 TB). The access to the data is served by five front end servers.

• Virtual organizations have here data in backup and archive policy.
• DU5 has migration policy disk_only
• Fotos

Object storage CL1 is based on Ceph technology. It is consisting of 17 storage servers, each has 48pcs HDD with total capacity 10.56 PB. The access to the data is served by three front end servers.

• available services: RBD a S3.
• Photos of the Data Storage

Object storage CL2 is based on Ceph technology. It is consisting of 33 storage servers, each has 48pcs HDD (each 16 TB) with total capacity 25.43 PB. The access to the data is served by five front end servers.

• available services: RBD a S3.
• Photos of the Data Storage

Object storage CL3 is based on Ceph technology. It is consisting of 74 storage servers, each has 24pcs HDD with total capacity 33.67 PB. The access to the data is served by five front end servers.

• available services: RBD a S3.
• Photos of the Data Storage

Object storage CL4 is based on Ceph technology. It is consisting of 57 storage servers, each has 24pcs HDD with total capacity 26.607 PB. The access to the data is served by five front end servers and nine aplications servers.

• available services: RBD a S3.

Object storage CL3 is based on Ceph technology. It is consisting of 50 storage servers, each has 30pcs HDD with total capacity 25.536 PB. The access to the data is served by five front end servers.

• available services: RBD a S3.
• Photos of the Data Storage

The differences between individual data centers are in the use of different data storage technologies. The choice of storage is initially determined by the desired service and type of data handling. The administrators of the Data Storage Department of CESNET take care of these requirements.

The services provided also include system security. The main emphasis is primarily on cyber security, which has three levels of security: Protection against DDoS attacks, network traffic security with a firewall, web application firewall.

Repository history