Solid State Storage in the Evolution of the Data Center Trends and Opportunities Bruce Moxon CTO, Systems and Solutions stec Presented at the Lazard Capital Markets Solid State Storage Day New York, June 12, 2013
SAFE HARBOR STATEMENT This presentation contains forward-looking statements, based on current management expectations. Any statement that refers to expectations, projections or other characterizations of future events or circumstances is a forward-looking statement, including, but not limited to, those relating to stec s product line key differentiators, objectives and innovation; the benefits, capabilities, performance, cost-savings and energy efficiencies of stec s products, solutions and other developing technologies; the rapidly evolving enterprise-storage and server markets; the size and strength of these market; the growing adoption and qualification of solid-date drives (SSDs); and the adoption of SSDs in general into new markets and applications. Because forward-looking statements relate to the future, they are subject to inherent risks and uncertainties, including factors that could delay, divert or change any of them, and could cause actual results to differ materially from those expressed, including due to the risk factors described in stec s periodic filings made with Securities and Exchange Commission. stec cautions you therefore to not place undue reliance on such statements. stec undertakes no obligation to update these forward-looking statements or relevant risks, except as may be required by law. 2
Enterprise Data Center Evolution 3
Scale-out, HPC, and HyperScale Data Centers Web Browsers Client Internet / Intranet Load Balancers Web / App Servers Service Application Services Caching (e.g. memcached) Resource Analytics Master DBs Slave DBs 4
Cloud Architectures and SAN Disaggregation Traditional and Virtualized Data Centers Cloud Architecture Data Centers Monolithic SANs 5
Why Open Compute Matters for more than Facebook Traditional Infrastructure Lifecycle Hyperscale Infrastructure Lifecycle 6
Software-defined Storage Unified Storage Appliances Virtual Storage Appliances (VSAs) Converged IO Big Data Storage Commodity DAS Hardware Storage Software Software-defined Storage On-premise Cloud Gateways Bare Metal (Storage Appliance) Virtual Storage Appliance (VSA) 7
Software-defined Storage Unified Storage Appliances NexentaStor (ZFS), Windows Storage Server 2012 Virtual Storage Appliances (VSAs) and Converged IO Modular Midrange vendor VSAs (NetApp, EMC) Virsto, DataCore SANSymphony-V ScaleIO, Converge.io, Exablox RedHat/Gluster Big Data and HPC Storage HDFS, NoSQL and K-V stores Lustre, IBM GPFS (SNC), StorNext Standardized Configurations & Integrated Appliances Cloud Storage Ephemeral (image) and Block Storage Services (SAN/NAS): OpenStack Glance and Cider (Nexenta, Ceph, ) Object Storage: Amplidata, CleverSafe, OpenStack Swift, Amazon S3* (Nexenta, Ceph, ) On-premise Cloud Gateways:TwinStrata, Amazon Storage Gateway*, Nirvanix* 8
Solid State Performance Characteristics (General) HDD (SAS) Sequential Random Read 200 MB/s 200 IOPS Write 200 MB/s 200 IOPS Rand Response 2-8x 100-1000x SSD / PCIe Sequential 4K Random Read.5 1.5 GB/s 50-300K IOPS Write.3 1 GB/s 20-80K IOPS Read Write HDD 8 ms 0.5 ms (cached*) SSD 60 us 20 us General Performance Characteristics. YMMV, depending on Device architecture Interface (SATA, SAS, PCIe) 2-4x performance range Transfer sizes QDs (concurrency) Cost Differential $0.50 - $1 / GB - SAS HDD $1 - $12 / GB - SSD/PCIe Sweet Spot High Random IOPS (esp. Read) Low Latency 9
Solid State Deployment Options High Performance, Low Latency DAS Servers SAN / DAS Cache (incl. Controllers) Alternative Flash Access Methods Shared Network Caching (NAS / SAN) Network Controller Cache Tiered Storage (Hybrid LUNs) Storage Systems High Performance Storage Pools All Solid State Arrays 10
Predominant Use Cases Use Case Transactional Databases Server Virtualization / Cloud VDI BI / DW / Analytics (including Hadoop, NoSQL, in memory analytics) Media High Performance Computing Benefit Improved performance, reduced query latency; virtualization/consolidation enabler Increased consolidation factors and per-server performance Reduced infrastructure cost ($/VDI); boot storm elimination Decreased query response time; increased concurrency and throughput Realtime access to high demand shared assets Accelerating a wide range of I/O-intensive HPC applications Common Theme: Increased Performance Density Increase in Consolidation Factors Reduction in Performance Sprawl 11
Deploying stec Solutions High Performance, Low Latency DAS Servers SAN / DAS Cache (incl. Controllers) s600 SATA s800 SAS s1100 PCIe Alternative Flash Access Methods EnhanceIO FOS Shared Network Caching (NAS / SAN) Network Controller Cache Tiered Storage (Hybrid LUNs) Storage Systems High Performance Storage Pools ZeusRAM s1100 PCIe All Solid s800 State SAS Arrays Reference Architectures 12
Revenue ($M) SSD Market Trends Enterprise SSD Market Revenue $6,000.00 $5,000.00 $4,000.00 $3,000.00 $2,000.00 $1,000.00 $0.00 2011 2012 2013 2014 2015 2016 Other PCIe SCSI SATA SAS FC 4% 21% 27% 13% 4% 31% External Storage Internal Storage (3 or more drives) Cloud (Direct or ODM Purchase) Cache Low End/Blade Server Workstation Market Direction: Bulk of revenue is in SAS and PCIe. SATA is expected to be steady revenue, cost driven and targeted for Cloud, blade and workstation segments *Source: IDC 13
S3000 Solid State Storage Solution Unified storage solution Block: iscsi File: SMB 2.1 / 3.0, NFS v3 / v4.1 Active-active continuous availability Up to 96 TB SSD storage High Availability configuration Fully redundant, multi-pathed architecture Redundant hot-swap power supplies and fans Enterprise features De-duplication, Thin provisioning, snapshots SMB3 direct, multi-channel Clustering Supports up to 8 nodes using Clustered Shared Volumes Simple/ familiar Windows user interface High performance 1.2M / 600K IOPS random 4K (SMB3) 8 / 5.5 GB/s sequential 1M (SMB3) 14
Host-side Caching Oracle Latency Reduction SQL Server SharePoint Exchange SAN (FC, iscsi) RAID Array (SAS) MySQL Virtualization Accelerates Applications Extends the Capabilities of the SAN 15
Flash 3.0 Flash technology presents new opportunities for application optimizations in the data path Random access media Near-memory latency at near-disk cost These will be optimally delivered through new device abstractions better-matched to device characteristics Memory abstractions Object abstractions Flash Object Store Library File System OS Buffer Cache Block Device Application Library Flash-optimized Device Flash Management Natural extension to Storage Class Memory 16
Thank You!