A Brief Description of the Botany Storage Cluster
a) The Botany Storage Cluster is a jointly funded storage solution not only to accommodate our current needs, but also to provide a reliable and scalable foundation for our growing storage requirements in the future.
b) This service is built with Microsoft Storage Spaces Technology as the software solution, along with commodity hardware for best value.
c) As the minimum system to start, there are two Dell servers (loaded with Windows Server 2012 R2; they function as fail-over controller nodes), both of which are connected to one SuperMicro JBOD chassis (with 45 SAS hard drive slots). When fully populated with 4TB SAS hard drives, the minimum system can provide 180TB of raw storage space. In the future, it will be possible to scale the system out to 4 JBOD chassis and provide up to 720TB of raw storage space (or more when using larger hard drives).
d) For better redundancy and data integrity, this service has several built-in features at different levels:
- There are two controller nodes that are configured as a failover cluster. When one node fails, the other node will take over all functionality within seconds.
- Each hard drive is connected to both controller nodes through dual channels on each SAS expander card, to avoid a single point of failure.
- In each disk pool, multiple hard drives are configured with parity settings (similar to RAID5) to avoid data loss under a single hard drive failure situation.
- When formatted using the ReFS file system, the system can monitor the data integrity in a particular partition to guard against data degradation in the storage media, and to automatically recover the original data if decay is detected.
e) To use this service, the customer will need to purchase one or more SAS hard drives through Botany IT according for their storage space. Botany IT will recommend the best hard drive model at the time of procurement. The total usable space for a customer will be the sum of the capacities of all SAS hard drives purchased by the customer. Botany IT staff will help the customer to create a space allocation plan according to his or her specific needs.
f) We CANNOT use SATA hard drives for this system. We may accept used SAS hard drives provided by the customer, if:
- The SAS hard drive is 1TB or bigger;
- The SAS hard drive is an Enterprise class product by one of the major hard drive manufacturers, such as Seagate and Western Digital.
- The SAS hard drive is a Near Line SAS hard drive or faster.
- The SAS hard drive passes the hardware test by Botany IT.
g) The system supports multiple protocols to mount the storage space on the client systems:
- SMB shared folder. This is the recommended method for most users. During testing, we successfully mounted shared folders to Windows, Mac and Linux computers. Although the newest SMB3 protocol is more reliable and delivers much better performance, we still support SMB2 and SMB1 to accommodate older client systems.
- iSCSI target. It is possible to mount a virtual disk on the client system through the iSCSI protocol. This method actually delivers best performance in both sequencing and random Input/Output tests. However there is a 64TB space limit for each mountable virtual disk, and the iSCSI target cannot be mounted to more than one client system at a time.
- NFS shared folder. Although it is possible to provide NFS shared folders to our customers, we highly recommend our customers to use SMB instead since it will work with most systems and has delivered better performance during our tests. We will only provide NFS shared folders if you cannot use the other two methods.
h) The customer can also choose to mount A partial AMOUNT or all of the purchased storage space to Botany backup server and use it for backup purposes. The customer will get one free backup license (can be used to back up a maximum of 4 computers) if he or she chooses to do so.
i) Please note that the design goals of this storage cluster are to provide reliable, scalable, high capacity near line storage. The performance of this cluster won’t be good enough to function as online storage space for a High Performance Computing cluster.
Please note that all parts of this storage cluster is in one server rack, connecting to two UPS units and two breakers in the same emergency power panel back by the building emergency generator. However, there is no geographic redundancy for this system. Please talk with Botany IT staff if you require greater availability.
Eligibility and Cost
a) Since this is a jointly funded system, all Botany professors, associates and/or adjunction professors, teaching faculty members, and Botany sub-unit directors/managers can purchase their share of storage space through Botany IT with the accounts that they have signing authority to use.
b) Some use cases may be paid with the Botany General Purpose Operation fund, if approved by Botany IT Advisory Board.
c) The total usable space for a customer is roughly the sum of all disk space the customer will purchase, but may vary due to parity setting. The cost of the SAS near line disk will vary due to capacity and market change. In June 2014, a 4TB enterprise class near line SAS hard drive cost around $400 before tax. Botany IT will provide a suggestion for best value at the time of procurement.
Service Level Agreement: Botany_Storage_Cluster_SLA_June2014(revised)
User instruction: please contact Botany IT for this document