Skip to main content
Pure Technical Services

Pure as-a-Service Catalog

Currently viewing public documentation. Please login to access the full scope of documentation.

Block Storage Service

//Block Service
  Ultra Premium Performance Capacity
Description
Ultra block storage service for mission-critical, I/O intensive, very-low latency workloads. The underlying Service Infrastructure deployed is designed to deliver 2000IOPS/TiB under the following workload characteristics: 70/30 read/write; DRR = 4:1; block size = 16k; and concurrency = 60% and without any replication configurations enabled. Premium block storage service for mission-critical, low-latency workloads. The underlying Service Infrastructure deployed is designed to deliver 1000IOPS/TiB under the following workload characteristics: 70/30 read/write; DRR = 4:1; block size = 16k; and concurrency = 60% and without any replication configurations enabled. Cost-effective general performance block storage service for mission-critical low-latency or throughput workloads. Cost-effective capacity-optimized block storage service for Tier2 workloads.
Performance
Latency
sub-ms sub-ms sub-ms 3-5 ms
Minimum IOPS/TiB*

(effective used)

2000 1000  300  100 
Maximum IOPS/TiB*

(effective used)

7000 4000 3000 400
Max Read Performance / TiB
368 MB/s 245 MB/s 184 MB/s 16 MB/s
Max Write Performance / TiB
133 MB/s 92 MB/s 51 MB/s 7 MB/s
Pre-Reduced or Encrypted (PRE) Charge
PRE rates are 2x of the standard rate. PRE rates applies if DRR<2:1 and if Pure has to ship additional capacity due to PRE Refer to EULA and PaaS program description.  PRE rates are 2x of the standard rate. PRE rates applies if DRR<2:1 and if Pure has to ship additional capacity due to PRE Refer to EULA and PaaS program description. PRE rates are 2x of the standard rate. PRE rates applies if DRR<2:1 and if Pure has to ship additional capacity due to PRE Refer to EULA and PaaS program description. N/A
Suggested Use Cases
In Memory DB SAP HANA, Oracle, VoltDB, Couchbase. Ideal for a broad range of //Block performance workloads with ultra IOPS/TiB requirements. Ideal for a broad range of //Block performance workloads with premium IOPS/TiB requirements.

Ideal for a broad range of Tier1 workloads: Cassandra, MongoDB, MariaDB. VDI, Vmware, Oracle, Microsoft SQL Server, SAP, HyperV, Epic. Hybrid and multi cloud, contrainers, media libraries, short term storage, Migrate to cloud, DR to cloud, test/dev.

Tier2 Database and virtual machines, archive stores, test/dev, snapshot consolidation/retention. 


* Ultra-performance block and high-performance block services and underlying storage infrastructure are designed to deliver 2000IOPS/TiB and 1000IOPS/TiB under certain workload characteristics, while the performance block does not have any IOPS/TiB guidelines. The workload characteristics are 70/30 read/write, DRR=4:1, block size=16k, and concurrency=60%. However this is not a run-time metric and actual IOPS/TiB may vary based on the characteristics of the workload. Pure only evaluates underlying equipment based on the assumed workload characteristics. The performance can be achieved when used with block workloads only. Quote numbers assume FC (or NVME-OF) "DMA" Directo Memory Addressing performance from the host. iSCSI numbers will be lower.  

The service offers tremendous burst performance headroom and most customer workloads will be able to drive significantly more performance.  Pure as-a-Service can also support custom performance requirements that are higher than what is quoted in the service catalog.

Unified Fast File and Object (UFFO) Service

//UFFO Service
  Ultra Premium
Description
UFFO scale-out file and fast object storage service that is designed to deliver ultra-fast read and write. UFFO scale-out file and fast object storage service that is designed to deliver premium fast read and write.
Performance
Min Operations / TiB **

(effective used)

2075 920
Max Read Performance / TiB
150 MB/s  55 MB/s
Max Write Performance / TiB
60 MB/s  18 MB/s
Pre-Reduced or Encrypted (PRE) Charge
N/A N/A
Suggested Use Cases

Rapid Restore, Artificial Intelligence, Analytics, EDA, HPC, Software development, Code & artifact repository. 

Rapid Restore, Artificial Intelligence, Analytics, EDA, HPC, Software development, Code & artifact repository. Custom and COTS (Splunk, Vertica EON) S3 app support Rapid Restore (CommVault, Veeam). 


** I/O TiB numbers for Scaleout File Services used 100% 4K reads w/ 3.3:1 Compression. Assumes 100% concurrency. Real world real IOP/TiB numbers could be much higher because in theory, the performance of the entire FlashBlade system is available to any given workload. 

The service offers tremendous burst performance headroom and most customer workloads will be able to drive significantly more performance.   Pure as-a-Service can also support custom performance requirements that are higher than what is quoted in the service catalog.