REFERENCE ARCHITECTURE 1 Hedvig Distributed Storage Platform with Cisco UCS Solution Overview and Reference Architecture Elastic Scale to petabytes of data Start with as few as two nodes and scale to thousands. Add capacity if and when needed. Embrace the economics of industry-standard x86 infrastructure to build your storage. Software-defined storage is a key pillar of the software-defined data center. It enables organizations to better manage the explosive growth of data and deliver greater provisioning flexibility while also lowering costs. The Hedvig Distributed Storage Platform, a modern, software-based storage solution, combined with Cisco UCS servers delivers these benefits for enterprise compute environments running at any scale. Hedvig & Cisco software-defined storage for modern business Modern business requires new levels of speed and agility to meet the demands of a dynamic, fast-changing marketplace. The most successful companies Simple Provides block, file, and object storage Collapse monolithic, disparate storage solutions into a single, modern platform. Streamline and automate provisioning. Bring the simplicity of public clouds to your datacenter. rise to the challenge by investing in technology that enables rapid response to business opportunity. The software-defined data center (SDDC) is now the bedrock of modern IT, delivering the flexibility needed for rapidly changing data, application, and user requirements. Together Hedvig and Cisco deliver a cost-optimized software-defined storage solution that gives enterprises the ability to quickly meet new requirements and more effectively manage storage growth. The combination of Hedvig software and Cisco UCS server hardware provides a highly scalable, resilient, elastic storage system with the simplicity of cloud, and a complete set of enterprise capabilities. Whether you need to optimize storage for traditional applications, build a storage foundation for service-oriented architectures and on-demand IT, or both, Hedvig and Cisco can help. Hedvig and Cisco provide your business with key advantages including: Flexible Meets rapidly changing data and application requirements Ensure any application or OS benefits from modern storage. Evolve to software-defined, hybrid cloud infrastructure at the pace your business demands. Provision storage with the speed and simplicity of the cloud Dynamically expand capacity and performance Cut costs Lower storage TCO up to 60% Eliminate the headaches of traditional storage operations and maintenance
REFERENCE ARCHITECTURE 2 The Hedvig Distributed Storage Platform The Hedvig Distributed Storage Platform is a highly scalable software-based solution designed to transform how you deliver storage to your enterprise. Hedvig software deploys on-premises and in public clouds, creating an implicitly hybrid storage system. Its patented distributed systems engine powers elastic, flexible storage with the capabilities you expect from an enterprise-class solution. Hedvig provides a single, unified storage solution with support for block, file, and object protocols that predictably and dynamically scales storage performance and capacity with Cisco UCS Servers. The core components of the Hedvig Distributed Storage Platform are: Hedvig Storage Service: Forms a scale-out storage cluster with Cisco UCS Servers, aggregating spinning disk and flash storage resources into a virtualized storage tier. Hedvig Storage Proxy: Enables OS, hypervisor, and application access to the storage cluster via industry-standard protocols and provides client-side services including caching and deduplication. Hedvig APIs: Enable REST and RPC-based access to all Hedvig storage features for software-developers who wish to automate provisioning and management with self-service portals, applications, and clouds. Hyperscale and Hyperconverged Hedvig software can be configured in two modes, hyperscale, where storage capacity is scaled independently from application compute resources, and hyperconverged, where storage and application compute are bundled and scaled together. With Hedvig, a single storage system can incorporate both hyperscale and hyperconverged resources affording the freedom to choose deployment options that best fit the requirements of your workload. Hyperscale Scale compute and storage independently Hyperconverged Scale compute and storage together Once deployed, you simply provision virtual disks each customizable by protocol, capacity, protection, and a range of storage services for a perfect fit with your applications.
REFERENCE ARCHITECTURE 3 The Cisco Unified Computing System is designed to deliver: Reduced TCO at the platform, site, and organizational levels Increased IT staff productivity and business agility through justin-time provisioning and mobility support for both virtualized and non-virtualized environments A cohesive, integrated system that is managed, serviced, and tested as a whole Cisco UCS Servers The Cisco Unified Computing System (UCS) is a next-generation data center platform that unites compute, network, storage access, and virtualization into a cohesive system designed to reduce total cost of ownership (TCO) and increase business agility. Cisco UCS servers adapt to meet rapidly changing business requirements, including just-in-time deployment of new computing resources to meet requirements and improve business outcomes. With Cisco UCS, you can tune your environment to support the unique needs of each application while powering all your server workloads on a centrally managed, highly scalable system. Scalability through a design for hundreds of discrete servers and thousands of virtual machines, and the capability to scale I/O bandwidth to match demand Industry standards supported by a partner ecosystem of industry leaders Cisco UCS Integrated Infrastructure Solutions speed up IT operations today and create the modern technology foundation you need for initiatives like private cloud, big data, and desktop virtualization. Cisco UCS Director provides centralized automation of your physical and virtual resources, helping you take full advantage of our open ecosystem approach. Cisco UCS brings the flexibility of virtualized systems to the physical world in a way no other server architecture can, lowering costs and improving your ROI. Cisco offers various server models, each with varying configurations of processing power, memory capacity, I/O and internal storage capacities in order to address varying software-defined storage requirements.
REFERENCE ARCHITECTURE 4 Use Cases Together, Hedvig and Cisco enable a more effective storage solution for traditional applications while also delivering an optimal platform for new, modern applications. Traditional Workloads Server Virtualization Hedvig software supports the widest breadth of hypervisors, operating systems, containers, and clouds. Hedvig and Cisco UCS provide a single, scale-out storage platform that supports a consistent, high-performance workflow for provisioning storage in large, highly virtualized environments. Backup Hedvig scales dynamically with Cisco UCS servers to provide an efficient platform for backup, archiving, business continuity, and disaster recovery. Built-in hybrid cloud support and storage efficiency features streamline on and off-site data protection with the ideal economics and flexibility for long-term secondary data retention. VDI Hedvig delivers fine-grained control of storage services to meet the unique demands of hosted virtual desktops. The software takes advantage of the latest high-performance flash storage options and the cost economics of Cisco UCS to ensure the success of your VDI project. New Workloads Production Clouds Hedvig s cloud-like storage provisioning delivers simplicity for infrastructure-as-a-service (IaaS). It plugs seamlessly into cloud orchestration and service catalog tools like OpenStack, and supports container technologies like Docker deployed with microservices environments. Test/Dev Clouds Hedvig offers unparalleled simplicity and flexibility for storage provisioning in test/dev environments. By providing point-and-click provisioning with granular virtual disk policy selection as well as instant cloning, Hedvig makes it easy to deploy storage for development followed by a seamless move to production. Big Data Hedvig and Cisco provide the ideal storage platform for big data. The flexible storage architecture is a perfect match for the elastic nature of Hadoop and NoSQL. Hedvig s tunable replication, built-in compression and deduplication enable organizations to virtualize big data applications and build a highly efficient data lake atop a single, unified storage solution.
REFERENCE ARCHITECTURE 5 Solution Validation Hedvig and Cisco tested and validated the Hedvig Distributed Storage Platform software running in a hyperscale configuration with Cisco UCS C240 M4SX Series Rack-Mount Servers. Testing focused on system functionality and compliance with the open standards followed by Cisco Unified Computing Systems and Nexus. By proving compliance with a set of mandatory requirements, the solution validation provides assurance that the joint solution performs and interoperates as expected. Specifically, validation verified that Hedvig software successfully installs and interoperates with the Cisco UCS server platform, enabling a cluster of Cisco UCS C240 servers to function as a single highly scalable, elastic and resilient enterprise storage solution with advanced data protection, backup and disaster recovery capabilities. Functional testing Testing of the Hedvig and Cisco UCS solution successfully validated a range of operations and functionality. Setup and installation Testing verifies that the solution adheres to installation and configuration best practices including: UCS Service Profile configuration Installation of the base system OS Installation of VMware ESXi and vcenter Installation of Hedvig storage service software Installation of the Hedvig storage proxy Installation of the Hedvig deploy server Verification of IP addressing including forward and reverse DNS operations System operation and functionality Testing verifies that core storage operations are functioning and available: Scaling: Scale storage capacity with addition of new Storage Nodes without I/O interruption Block disk provisioning: create and mount an iscsi virtual disk NFS disk provisioning: create and mount an NFS virtual disk Compression: create and store data to virtual disks with compression enabled Deduplication: create and store data to virtual disks with deduplication enabled Dedupe results reporting: reflect space savings in user interface
REFERENCE ARCHITECTURE 6 Resilience testing Testing of the Hedvig and Cisco UCS solution successfully validated system resiliency, verifying continued operations despite component and system-level failures including: Hard drive failures: I/O continues as normal and data rebuilds to alternate drive(s) Disk replacement: I/O continues as normal during failed disk replacement Storage Node failure: I/O continues as normal and data rebuilds to alternate Storage Node(s) Storage Node network interface failure: I/O continues as normal on surviving nodes Storage Node reboot: I/O continues via alternate Storage Node Storage Proxy failure: I/O continues at client via surviving Storage Proxy Storage Proxy network interface failure: I/O continues via surviving Storage Proxy Storage Proxy reboot: I/O continues via alternate Storage Proxy Test configuration System topology The following graphic depicts the test environment: Figure 1. Hedvig and Cisco test environment
REFERENCE ARCHITECTURE 7 Host setup The following table lists details of the hosts used during system testing: Name Storage VLAN Internal Private VLAN STORAGE NODES eth0 eth1 hedvigdata1.cdnivt.cisco.com 10.0.107.41 hedvigdata2.cdnivt.cisco.com 10.0.107.42 hedvigdata3.cdnivt.cisco.com 10.0.107.43 STORAGE PROXIES (HA PAIRS) hedvigcvm1a.cdnivt.cisco.com 10.0.107.45 172.22.22.20 hedvigcvm1b.cdnivt.cisco.com 10.0.107.46 172.22.22.21 hedvigcvm1vip.cdnivt.cisco.com 10.0.107.47 172.22.22.2 hedvigcvm3a.cdnivt.cisco.com 10.0.107.68 172.22.22.24 hedvigcvm3b.cdnivt.cisco.com 10.0.107.69 172.22.22.25 hedvigcvm3vip.cdnivt.cisco.com 10.0.107.70 172.22.22.2 ESX HOSTS vmk2 esx1-vmkernel 172.22.22.1 esx2-vmkernel 172.22.22.1 LINUX VIRTUAL MACHINES centos-1.cdnivt.cisco.com 10.0.107.48 centos-2.cdnivt.cisco.com 10.0.107.49 centos-3.cdnivt.cisco.com 10.0.107.50 centos-4.cdnivt.cisco.com 10.0.107.51 SOFTWARE DEPLOYMENT SERVER hedvigdeploy1.cdnivt.cisco.com 10.0.107.44 Network infrastructure The following table lists details of the networking hardware used during system testing: Model Type Version UCS 6248UP Fabric-interconnect 5.2(3)N2(2.23d) UCS 6248UP Fabric-interconnect 5.2(3)N2(2.23d) Nexus 5548UP DC Switch Nexus 5548UP DC Switch UCS 2204XP Fabric extenders 2.2(3d) UCS 2204XP Fabric extenders 2.2(3d)
REFERENCE ARCHITECTURE 8 UCS Hardware The following tables list details of the UCS hardware used during system testing: Model/Type Host Name IP Address Description B200 M3 UCS3-C1-S3-102-22 (esx 2) 10.0.102.22 ESXi 5.5 B420 M3 UCS3-C1-S3-102-24 (esx 1) 10.0.102.24 ESXi 5.5 B200 M3 UCS3-C2-S3-102-26 (esx 3) 10.0.102.26 ESXi 5.5 C240 M3 hedvigdata1.cdnivt.cisco.com 10.0.107.41 Hedvig v1.0.2/centos 6.7 C240 M3 hedvigdata2.cdnivt.cisco.com 10.0.107.42 Hedvig v1.0.2/centos 6.7 C240 M3 hedvigdata3.cdnivt.cisco.com 10.0.107.43 Hedvig v1.0.2/centos 6.7 Model/Type Processor Socket / Cores per Socket Memory RAID Controller HDD/ SSD NIC Adapters B200 M3 1 / 10 64 GB 2x1TB ESX boot disk RAID 1 UCS VIC 1240 4.0(1d) B420 M3 2 / 6 128 GB 2x1TB ESX boot disk RAID 1 UCS VIC 1240 4.0(1d) B200 M3 1 / 10 64 GB 2x1TB ESX boot disk RAID 1 UCS VIC 1240 4.0(1d) C240 M3 2 / 4 64 GB C240 M3 2 / 4 64 GB C240 M3 2 / 4 64 GB LSI 6G MegaRAID SAS 9266-8i (23.29.0-0014) 2x1TB Linux boot disk RAID1 2x280GB SSD JBOD 18x1TB HDD JBOD LSI 6G MegaRAID SAS 9266-8i (23.29.0-0014) 2x1TB Linux boot disk RAID1 2x280GB SSD JBOD 18x1TB HDD JBOD LSI 6G MegaRAID SAS 9266-8i (23.29.0-0014) 2x1TB Linux boot disk RAID1 2x280GB SSD JBOD 18x1TB HDD JBOD UCS VIC 1225 4.0(1d) UCS VIC 1225 4.0(1d) UCS VIC 1225 4.0(1d)
REFERENCE ARCHITECTURE 9 Figure 2. Cisco UCS Manager - test environment topology Figure 3. Cisco UCS Manager - Storage Node details
REFERENCE ARCHITECTURE 10 Figure 4. Hedvig user interface showing 3-node UCS-based storage cluster Figure 5. Hedvig user interface showing 3-node UCS-based storage cluster
REFERENCE ARCHITECTURE 11 Installed components The following graphics depict the installed hardware, operating system, and software components of the test environment as well as the IP and networking configuration: Figure 6. Installed software and disk components Figure 7. IP and networking topology
REFERENCE ARCHITECTURE 12 Figure 8. VMware vsphere view of the test environment Software versions tested The following table lists details of OS and software versions tested: Component Quantity Software Version Hedvig storage software 3 1.0.2 Hedvig proxy 6 1.0.2 Hedvig deploy server 1 1.0.2 ESXi 3 5.5 Linux CentOS 5 6.7 Tested storage protocols The following table lists details of the storage protocols tested: Protocol Version Function iscsi RFC-3720,3721 Block data storage NFS 3.0 File storage
REFERENCE ARCHITECTURE 13 Recommended Configurations The following table provides recommended Cisco UCS hardware configurations based on three common Hedvig Storage Node profiles performance optimized to deliver high storage IOPS performance, capacity optimized to deliver a balance of capacity and mid-range performance, and dense storage to deliver high-capacity, deep storage requirements at best-effort performance. Storage Node Recommended Specifications Performance optimized Capacity optimized Dense storage Server Model Cisco C240 M4 Cisco C240 M4 Cisco UCS C3160 CPU Dual E5-2600v3 series CPU 8 cores per processor Dual E5-2600v3 series CPU 4 cores per processor Memory 128 GB 64 GB 128GB Disk Controller Network Disks for OS Disks for metadata & cache 2 Standard JBOD or RAID disk controller(s) 1 x 10GbE (2 x 10GbE for redundancy) (2) x 200GB boot disks RAID 1 for Linux (6) x 400GB/800GB SSD (JBOD 1 ) Disks for data 3 (18) x 1TB/2TB SAS/SATA 2.5" Hard Drives (JBOD 1 ) Standard JBOD or RAID disk controller(s) 1 x 10GbE (2 x 10GbE for redundancy) (2) x 200GB boot disks RAID 1 for Linux (3) x 400GB/800GB SSD (JBOD 1 ) (9) x 1TB/2TB/4TB/6TB SATA 3.5" Hard Drives (JBOD 1 ) Total disk slots 26 14 62 Dual E5-2600v3 series CPU 12 cores per processor Standard JBOD or RAID disk controller(s) 1 x 10GbE (2 x 10GbE for redundancy) (2) x 200GB boot disks RAID 1 for Linux (8) x 400GB/800GB SSD (JBOD 1 ) (52) x 1TB/2TB/4TB/6TB 3.5" SATA Hard Drives (JBOD 1 ) The following table provides recommended resource configuration for the Hedvig Storage Proxy. Storage Proxy Recommended Specifications High-performance clients / applications Mid-performance clients / applications vcpu 6 vcpu 4 vcpu 4 vcpu Memory 12 GB 8 GB 8 GB Disk (2) x 400GB/800GB SSD local disks 4 Low-performance clients / applications (1) x 400GB/800GB SSD local disks 4 No local SSD disks 4 1 For metadata and data use JBOD configuration. If JBOD not available, use RAID 0 for each individual disk. 2 Hedvig metadata and cache recommended to be placed on SSD. SSD to be 5%-7% of storage capacity on each node. More SSD required if pinning data to SSD. 3 Data disks configured in multiples of three. 4 Used as cache to improve performance of read I/O and deduplication highly recommended for clients that require low latency read I/O and high deduplication performance. Optional for low performance clients.
REFERENCE ARCHITECTURE 14 Summary: Hedvig & Cisco The product is easy to use and it can grow with us. We just replace or add servers as we extend capacity and update the system in the future. Hedvig s approach is more predictable and leverages our investment in Cisco, which means a more predictable cost. Christoffer Niemi, IT Architect LKAB The Servers provides a flexible software-enabled system with the capabilities required to support any application, hypervisor, container, or cloud. The validated solution provides a single platform powering in-software provisioning of file, block, and object storage with the flexibility to span private and public clouds, creating an elastic, hybrid cluster that can scale to thousands of nodes. By aggregating and virtualizing Cisco UCS server and storage resources including flash and spinning disk, the Hedvig software enables a single, elastic storage system that enables today s enterprises to simply and cost-effectively tailor a storage to support any workload. Resources Visit the following links for additional product and solution information: Cisco UCS page http://www.cisco.com/go/unifiedcomputing Cisco Marketplace Hedvig page: https://marketplace.cisco.com/catalog/companies/hedvig-inc/products/ hedvig-distributed-storage-platform-for-cisco-ucs Hedvig Distributed Storage Platform product page: http://www.hedviginc.com/product Hedvig technical overview white paper: http://www.hedviginc.com/resources/product-technical-whitepaper Hedvig product demo video: http://www.hedviginc.com/resources/video-guided-product-tour LKAB customer case study: http://www.hedviginc.com/resources/case-study-lkab LEARN MORE ABOUT HEDVIG Built by software engineers of the world s largest distributed systems, Hedvig delivers modern storage for enterprise compute environments running at any scale. Customers using the Hedvig platform transform their storage from a box where data resides to a fundamental business enabler. 2016 Hedvig Inc. All rights reserved. Version 1.0 2350 Mission College Blvd, Suite 500 Santa Clara, CA 95054 www.hedviginc.com