COMPARISON OF VMware VSHPERE HA/FT vs stratus ftserver SYSTEMS
White Paper 2 Ensuring Availability of Virtualized Business-Critical Applications in an Always-On World Introduction Virtualization has become standard practice in organizations of all types and sizes around the world, due to the advantages it offers such as agility, efficiency, and scalability compared to traditional physical infrastructures. Initial concerns about virtualizing applications, including performance, support, and licensing, have disappeared thanks to advances provided by technologies such as VMware vsphere. In fact, organizations by the thousands have gained the confidence to virtualize even business-critical applications such as Microsoft Exchange, Oracle Database, and SAP Enterprise Resource Planning. As more and more business-critical applications are virtualized, however, a new concern has emerged: availability. Outages are costing companies more and more each year with the average cost of an hour of downtime for large companies now at a staggering $686,250 1. While some applications can tolerate brief outages, downtime of business-critical workloads, with the associated risk of data or transaction losses, is simply unacceptable. And with multiple virtual machines (VMs) running on a single physical host, one hardware failure can have widespread business impact. This paper presents two distinct approaches to ensuring availability for virtualized business-critical applications: VMware vsphere HA or FT and Stratus ftserver systems. The paper compares each solution in terms of installation, maintenance, and costs. Most important, it highlights the fundamental differences regarding downtime recovery versus downtime prevention. Solution Implementation: Complexity vs. Simplicity Establishing a high-availability environment based on clustering is often a complex and costly endeavor. VMware vsphere HA requires, at minimum, a two-node cluster to enable failover protection for VMware environments (Figure 1). Similarly, vsphere FT creates a shadow instance of a VM, which requires multiple hosts and additional storage configuration. vsphere FT also requires a dedicated network for VMware vmotion to enable failover. In a VMware environment comprised of many hosts, each node belonging to the HA/FT cluster must be identified and appropriate network connectivity between them established. The solution also requires an external storage array. It is then necessary to configure a number of settings to determine how the HA or FT cluster will behave and ensure that adequate resources are available in the event of a failover. In today s always-on world, ensuring availability of virtualized business-critical applications is absolutely essential. Most approaches to minimizing downtime, such as VMware vsphere High Availability (HA), employ server clusters and failover mechanisms that restart VMs on another host in the event of a hardware or operating system fault. However, the process of recovery not only takes time but also implies that damage has already been incurred. Ideally, the systems on which virtualized business-critical applications run should prevent downtime in the first place by working through system faults, thus avoiding business disruption entirely. VMWare Host 1 External LAN LAN 1 LAN 2 Heartbeat 1 Heartbeat 2 VMWare Host 2 LOGICAL SERVER Switch 1 Switch 2 SAN Figure 1: Two-node vsphere HA cluster.
White Paper 3 By contrast, Stratus ftserver systems provide out-of-the-box fault tolerance for VMware environments with no additional hardware, network, or software requirements (Figure 2). Virtual Machine 1 APP OS Virtual Machine 2 APP OS Virtualization Layer Virtual Machine 3 Figure 2. Stratus ftserver Fault-Tolerant System Stratus ftserver systems are built on cost-effective industrystandard hardware powered by Intel multi-core processors. Installation is fast and easy, offering full support for all standard vsphere products including Distributed Resource Scheduler (DRS), vmotion, Storage vmotion and HA with no additional configuration work or system modifications required. A typical vsphere HA or FT cluster could take several days to install, configure, and validate considering the requirements to provide multiple servers, dedicated cluster networking, and shared storage. In contrast, a Stratus ftserver system can typically be installed in a few hours. DB OS Total Cost Evaluation When evaluating the cost of ensuring availability using vsphere HA/FT clusters versus Stratus ftserver, it is important to look at the whole picture. A vsphere HA/FT cluster requires at least one additional server, as well as a management console, a high-availability network, and external storage. With the average life of a standard server just three to five years, recurring hardware expenses can also be expected along with downtime to perform migrations and upgrades. In addition, the cluster may require multiple software licenses for each server and the management console. Stratus ftserver systems are a complete, integrated solution for ensuring continuous application availability. They require no additional servers, networking, or storage. Moreover, only one license is needed for the operating system and application compared to multiple operating system licenses in a cluster scenario, lowering total costs substantially. In addition, ftserver systems have historically averaged eight years in service, enabling IT organizations to stretch their budgets and minimize upgrade cycles. Another consideration is the ongoing cost of renting VMs off a virtualized cluster infrastructure compared to having a dedicated standalone solution like ftserver. Many times, IT departments will set up these virtual machines and cross-charge to each department using them. The increased costs and IT skills needed to run a VMware cluster can be passed to each individual department. The cost-efficient design and simple skills needed to run an ftserver system make it a more economical choice. Management and support are also important cost considerations. Clusters require extensive hands-on care, often necessitating the expense of full-time administrative staff. In fact, cluster capacity, policies, resources, and software changes must all be regularly managed and tested to validate proper operation.
White Paper 4 With resilient hardware and built-in service technology, administration of ftserver systems is minimal. In fact, the Stratus automated service technology makes ftserver systems essentially self-managing, requiring little to no hands-on attention and, consequently, very little cost. When service is required, Stratus also performs remote software updates and proactively delivers hardware components on-site that can simply be plugged into the ftserver system while it continues to run. As a result, organizations can seamlessly upgrade the system without disrupting operations and avoid the normal IT lifecycle of rip and replace every few years. Over the long term, this ease of servicing avoids hundreds of thousands of dollars in upgrade expenses typically associated with clustered infrastructures. Figures 3 and 4 highlight typical cost differences using output from the Stratus Total Cost of Ownership Analyzer. Availability Report Card In terms of maximizing availability of business-critical virtualized applications, vsphere HA and ftserver systems offer two fundamentally different approaches: vsphere HA clusters rely on system failover while ftserver systems provide downtime prevention. Because of these differences, there are several important factors to consider when evaluating the two solutions. While vsphere HA can reduce application downtime by automatically restarting virtual machines upon VM failure, it does not protect the host server or hypervisor against downtime or performance degradation. However, this restart may take many minutes or hours for large systems. Moreover, any data uncommitted to disk is lost during application or server crashes and restarts (see Figure 5). TCO Summary (Three Years) Competitor Stratus Hardware $21,987 $13,324 Software $13,143 $16,880 Installation $10,198 $2,950 Support $9,592 $17,964 Administration $9,809 $3,462 Sub Total $64,729 $54,580 Downtime $13,104 $262 TOTAL TCO $77,833 $54,842 $100,000 $80,000 $60,000 $40,000 $20,000 $0 TCO = $77,833 VMware HA Cluster Hardware Software Installation TCO = $54,842 Stratus ft2710 w/vmware Support Administration Downtime Restart time varies depending on the application. Below are some observations: ABAP Central / Dialog Instances - ~4-5 mins for OS restart and SAP to come online. Java Central / Dialog Instances - ~15-17 mins for OS restart and SAP to come online. DB Instance - ~4-5 mins for OS restart and SQL Server to recover. Web Dispatcher Instance - ~3-4 mins. Source: SAP Saber Carving SAP into separate landscapes for company split. Van Vi and Rick Jones Figure 5: VMware vsphere HA restart times. Figure 3: TCO three year comparison for manufacturing application. TCO Summary (Five Years) Hardware Competitor $21,987 Stratus $13,324 $120,000 $100,000 TCO = $101,564 Software Installation $13,143 $10,198 $16,880 $2,950 $80,000 $60,000 TCO = $69,301 Stratus ftserver systems deliver fault tolerance and seamlessly integrate with standard vsphere products to avoid downtime completely instead of recovering from it. They provide 99.999+ percent availability to protect business-critical applications from host faults that would otherwise result in VM failures. And in-flight data for process and batch applications is fully protected from loss. Support $18,048 $29,940 Administration $16,348 $5,770 $40,000 $20,000 Sub Total $79,724 $68,864 Downtime $21,840 $437 TOTAL TCO $101,564 $69,301 Figure 4: TCO five year comparison for manufacturing application. $0 VMware HA Cluster Hardware Software Installation Stratus ft2710 w/vmware Support Administration Downtime Stratus ftserver systems deliver fault tolerance and seamlessly integrate with standard vsphere products to avoid downtime completely instead of recovering from it.
White Paper 5 Figure 6 provides a side-by-side comparison of the key availability capabilities offered by each solution. VMware HA Stratus ftserver Availability Philosophy Fast recovery after failure Failure prevention Typical Availability Level 99.9% to 99.99% 99.999% + Points of Failure CPU s & memory; storage array & initiators; switches None Susceptibility to Lost Data Any in-flight data is lost None Ability to Detect and Isolate Transient Errors No Yes Dependency on IT Staff Moderate to high Low Configuration Requirements 2 or more servers & OS; shared storage; may require multiple copies of software 1 server & OS Other Requirements Requires vcenter Server None Single System View Yes Yes Figure 6: VMware vsphere HA and Stratus ftserver availability matrix. In addition, VMware FT clusters have limitations. Currently, VMware FT supports only a single processor core (vcpu) per virtual machine, a restriction that significantly limits its applicability to protect most business-critical applications. Additionally, it supports a maximum of four FT-protected VMs per host. Typically, Tier 1 applications are performance intensive and require multi-core symmetric multiprocessing (SMP). For example, Microsoft recommends four to 12 dual-socket cores for Exchange and four to eight cores for SharePoint and SQL Server. Oracle recommends six to 12 cores. After trying for more than half a decade, VMware has recently announced plans to support up to four vcpus per VM as part of the 2015 release of vsphere 6.0. Stratus imposes no hard limits on the number of vcpus or amount of virtual memory per virtual machine. Furthermore, the number of VMs that can be hosted on an ftserver is limited only by its configuration and an organization s polices on system over-provisioning. VMware has been reporting that the performance limitations expected in their new VMware FT offering make it not appropriate for low-latency or high transaction volume applications, which is exactly where SMP fault tolerance is needed. Stratus ftserver systems have none of these performance limitations and are ideal for the most demanding mission-critical applications. Finally, Stratus is so confidant in the mainframe level of availability ftserver systems deliver with VMware, there is a $50,000 guarantee for customers reporting any unplanned downtime for both the server as well as vsphere. Conclusion With organizations of all types and sizes having greater reliance on virtualized business-critical applications, ensuring availability of the VMware environment is a top priority. VMware vsphere HA and FT provide the availability needed for certain applications but still put organizations at risk for periods of downtime and potential data loss. To strengthen availability of VMware environments, Stratus ftserver offers organizations a fault-tolerant solution that prevents downtime and protects against data loss. While vsphere HA reacts to an application or system failure by invoking recovery processes, ftserver systems continuously take proactive measures to prevent downtime from occurring in the first place. With vsphere HA, VMs must be restarted on another host meaning that in-flight data is lost. In contrast, ftserver systems ensure that no data is lost. Availability clusters, as used with vsphere HA and FT, are inherently more complex. Even when performing perfectly, dependencies among third-party application components can result in extended restart times. Stratus ftserver is a simple, easy-to-use, turnkey solution that allows organizations to run VMware environments out of the box and automatically gain reliable downtime-prevention with no external dependencies and no special IT skills needed to operate When comparing vsphere HA/FT and ftserver systems the differences are clear: vsphere HA clusters require complex implementation and administration, multiple software licenses, and still, data loss can occur during failover with the vsphere HA product. Stratus ftserver systems offer plug-and-play simplicity, single software image, and zero data loss. Ultimately, ftserver systems provide the highest possible availability for VMware environments for the lowest total investment in time and money.
White Paper 6 Find Out More About Stratus ftserver Systems For complete details on how Stratus ftserver systems prevent downtime and ensure continuous availability of business-critical virtualized applications, visit http:///products/platforms/ftserversystems About Stratus Stratus Technologies is the leading provider of infrastructurebased solutions that keep applications running continuously in today s always-on world. Stratus enables rapid deployment of always-on infrastructures from enterprise servers to clouds without any changes to applications. Stratus flexible solutions software, platform and services prevent downtime before it occurs and ensure uninterrupted performance of essential business operations. To learn more, visit stratus.com Stratus ftserver offers organizations a fault-tolerant solution that prevents downtime and protects against data loss. 1 Preventing Virtual Application Downtime, Jim Rapoza, Aberdeen Stratus ftserver systems provide the highest possible availability for VMware environments for the lowest total investment of time and money. Specifications and descriptions are summary in nature and subject to change without notice. Stratus and the Stratus Technologies logo are trademarks or registered trademarks of Stratus Technologies Bermuda Ltd. All other marks are the property of their respective owners. 2014 Stratus Technologies Bermuda Ltd. All rights reserved. 141017