Understanding Hyperconvegence Is It Right for Your Business?

DESIGN | HOST | MANAGE

Table of Contents Introduction

1

Chapter 1

2

What Is Converged Infrastructure?

Chapter 2 What Is Hyperconverged Infrastructure?

Chapter 3 Who Are the Major Players in the Hyperconverged Market?

Chapter 4 How Will HCI Impact My Infrastructure?

Chapter 5 Ways to Apply HCI

Conclusion

3

6

8

9 10

DESIGN | HOST | MANAGE

INTRODUCTION Integrated systems are combinations of server, storage, and network infrastructure. They are sold with comprehensive management software to operate and manage the complete system. These converged infrastructure systems are delivered pre-assembled and tested from the vendor, or they are built by local IT infrastructure firms according to vendor-prescribed blueprints. Available since 2010, the better known products in this market are the VBlock from VCE and FlexPod from NetApp. While they have been relatively successful in delivering on their promises, the cost has kept them out of the reach of most small and medium-size businesses. Hyperconverged (HC) systems are the latest evolution of integrated systems. They offer much tighter coupling of compute, network, and storage hardware resources, eliminate the need for a traditional storage area network (SAN), and can be built on commodity hardware. Because the HC market is still nascent, there are many questions in the minds of IT decision makers, such as: What is converged infrastructure? • Does it run out of the box or is some assembly required? If the latter, can my staff handle it? • Is it difficult to manage?  hat exactly is a hyperconverged system, and how is it different from a W converged system? • Will it save me money? • Do I still need my hypervisor and/or its management tools? • A  m I going to be locked into an expensive proprietary relationship with a vendor? Who are the top vendors? Must I replace my existing infrastructure? Where would I use it in my business and why? •

Will it help my company be more competitive?

Confronting Hyper-confusion If you are considering hyperconverged infrastructure, you probably have many questions. This eBook is designed to help you understand the basics and provide you with resources to dig deeper into this topic.

DSScorp.com

1

DESIGN | HOST | MANAGE

Chapter 1: What Is Converged Infrastructure? Gartner noted that the integrated systems market, which includes converged and hyperconverged infrastructure, is increasing by 50% each year. Converged infrastructure (CI) is simply another name for integrated systems. It is a combination of server, storage, and network infrastructure that is sold as a combined unit, along with software to manage it all. Converged infrastructure has emerged as an increasingly popular alternative to traditional siloed architectures. Gartner noted that the integrated systems market, which includes converged and hyperconverged infrastructure, is increasing by 50% each year.1 Traditionally, CI has consisted of a carefully designed integration of standard servers (often blade architecture), external storage in the form of a SAN, and top-of-rack network switches. It can be delivered as a pre-integrated and tested solution in a full or half rack, or it can be hand-assembled on your premises from individual components by a certified reseller. Examples include VCE Vblock, Lenovo Converged System (formerly PureFlex), and NetApp Flexpod. Regardless of how it is delivered, CI provides a virtualized shared infrastructure that can be quickly deployed and is suitable for almost any type of workload. While most CI systems are often sold as plug and play, the result varies greatly by vendor. In addition, the robustness of the management software determines ease-of-use. Many vendors offer a unified dashboard that covers everything from provisioning through backup and disaster recovery. Management and Support of CI CI lends itself to rapid deployment and a certain degree of simpler management. It can be expensive, though. A careful cost/benefits analysis will include the expected level of operational cost reductions. Ideally, highly skilled resources will spend less time maintaining the system or a lower skilled resource can act as a substitute. Support is simplified with a converged infrastructure. A single vendor is responsible for supporting the entire environment, right down to keeping the firmware levels in sync across all of the various components. It is unclear, however, how successful this support has been in terms of time saved and its impact on system availability. For example, do vendors have undue control over the scheduling and duration of downtime for maintenance? Answers to these sorts of questions are anecdotal at best, but one can conclude there must be some impact, and it requires consideration.

1

CRN | http://www.crn.com/news/data-center/300073165/gartner-converged-infrastructure-magic-quadrant-nutanix-top-visionary-vce-overall-leader.htm

DSScorp.com

2

DESIGN | HOST | MANAGE

Chapter 2: What Is Hyperconverged Infrastructure?

Converged infrastructure has evolved into hyperconverged infrastructure (HCI). Like CI, HCI integrates server, storage, and network infrastructure, but in a much more tightly coupled fashion. HCI eliminates the need for a traditional storage area network (SAN) in favor of using locally attached storage on most or all server nodes in the environment. The HCI management software provides all of the necessary storage management functions and some degree of compute provisioning.2 Besides basic storage functions, the software often provides additional capabilities like backup, recovery, replication, deduplication, and compression. Some vendors provide their management software as a standalone application, while others provide it as a plugin to the hypervisor’s existing management interface. Most require some sort of control VM that runs on every node in the cluster. HCI Is on the Rise The rise of HCI is part of a larger trend towards software-defined architecture and the software-defined data center. Management is accomplished at the virtual machine level rather than at the individual component levels of compute, storage, and networking. Instead of thinking in terms of CPU cores, GBs of memory, and GBs of disk, you can think in terms of various VM configurations. While the CI market is more firmly established, IDC sees the HCI market as still being in the early development and penetration stages with potential for growth.3 Cisco’s recent re-entry into the marketplace with yet another approach to HCI demonstrates that the technology is still changing rapidly.

By “some degree,” we refer to the fact that some vendors simply use the native hypervisor’s interface, such as vCenter, for provisioning rather than providing their own interface to automate some or all of the process. Storage Newsletter | http://www.storagenewsletter.com/rubriques/market-reportsresearch/idc-marketscape-worldwide-hyperconverged-systems/

2 3

DSScorp.com

3

DESIGN | HOST | MANAGE

A Word About Networking We have said that network infrastructure is part of the integration for CI and HCI, but it is probably the least integrated part of most vendors’ solutions. CI/HCI management software often does not have an awareness of the top-ofrack network switches or how to interact with them. For example, Cisco’s HyperFlex offering promises much more functionality in this area. However, experienced networking skills will be required to implement these solutions.

How Is Hyperconverged Infrastructure Different? Unlike CI, HCI can be deployed in smaller units, which leads to lower initial costs. Many HCI offerings are designed around a building block of a single physical enclosure, or appliance containing 4 servers or nodes. Each node has one or more CPU cores, memory, SSD, and spinning hard drives. The management software takes care of writing data redundantly across multiple drives and nodes to provide a high degree of fault tolerance. When multiple appliances are deployed, companies can achieve high degrees of availability. The SSD tier in every node provides fast performance. Some vendors are even able to recover a lost node automatically with no human intervention. HCI can be built on tier 1 server hardware (e.g. Cisco, HP) or on cheap commodity hardware. 1U or 2U boxes are the norm, but blade servers can also be used. Some HCI players sell appliances that are ready to run out of the box, while others simply provide software for implementing the storage management functions. Do You Still Need Your Hypervisor? Typically, you will need to provide a hypervisor, and this is another area to carefully consider. Not all HCI vendors support all hypervisors. They may support some better than others, or they may try to push their own proprietary hypervisor instead. If you have already deployed a virtualized environment, HCI can be easily implemented alongside a hypervisor without any negative impact.

DSScorp.com

4

DESIGN | HOST | MANAGE

Benefits of HCI • Scalability: Respond quickly to workload requirements by simply adding a compute or storage appliance. The more mature offerings grant short deployment cycles, providing greater flexibility and agility. • Optimization: Utilize HCI solutions to consolidate IT resources, as well as reduce staff and data center expenses. • Efficiency: Manage resources through a consolidated management interface to reduce operating costs and free administrative resources to focus on more strategic business activities. • Resiliency: Gain better resiliency within single sites and across multiple data centers and remote/branch offices as a result of the HCI design. Also, harness advantages over traditional environments and CI by natively providing data management functions, like deduplication, compression, backup, recovery, and replication. • Quality: Deliver an always on, always more cloud-like end user experience. HCI technology enables IT to deliver on expectations for service quality, timeliness, high availability, and disaster recovery.

Limitations of HCI Once you’ve selected a particular HCI technology, you are locked into that vendor.4 Also, the hypervisor you choose may affect the level of software-defined architecture capabilities available to you. Because of the extremely tight coupling, you may be hampered by your vendor when you wish to update software levels. For example, you may be unable to upgrade to a new level or even patch vSphere until the vendor supports it. While there are notable exceptions, HCI hardware components are typically offered in just a small number of configurations. This could become a limiting factor either technically or financially. In addition, not all vendors support the adding of just compute or just storage as your needs grow, which could present a financial constraint when scaling.

4

This is also true of CI technology.

DSScorp.com

5

DESIGN | HOST | MANAGE

Chapter 3: Who Are the Major Players in the Hyperconverged Market? As HCI gains in popularity, more vendors continue to join the market. The range of choices can be overwhelming, though. To simplify the decisionmaking process, here is a brief look at some of the market leaders, some of their differentiators, and the most recent innovations they have added to their HCI offerings. Nutanix Nutanix has been in the game since 2009, shipping its first product in 2011. With over 1,100 employees, Nutanix operates in more than 70 countries across 6 continents. Several of their 1,500 customers have installations in excess of 100 nodes, and one is reported to have 1,500 nodes. While supporting both VMware and Hyper-V, Nutanix also has developed its own hypervisor, Acropolis, based on KVM. The company maintains strategic alliances with VMware, Dell, Microsoft, and a number of other top software vendors. By eliminating the need for external SANS and building compression, deduplication, and erasure coding into its storage fabric, Nutanix offers low total cost of ownership without sacrificing availability, performance, and scalability.5 Their management software, Prism, is currently the most robust in the marketplace. Nutanix is one of the most mature offerings in today’s market. SimpliVity SimpliVity began shipping HCI products in 2013. Their OmniCube product family is based on commodity hardware. Their OmniStack solutions are available on Cisco UCS and Lenovo System x. SimpliVity’s solutions are differentiated by hardware-assisted compression, deduplication, and data protection. Scaling can be done in small increments, and the system is designed to do in-line deduplication and compression at origin. SimpliVity’s management tools connect to your hypervisor’s existing management framework (e.g. vCenter) using standard APIs. Prior to Cisco announcing HyperFlex, SimpliVity successfully ran on Cisco UCS. Now, however, the UCS future is a bit unclear.

5

Gartner Magic Quadrant for Integrated Systems, pg. 17, Doc ID: G00266749

DSScorp.com

6

DESIGN | HOST | MANAGE

Cisco Cisco has recently released HyperFlex, a hyperconverged infrastructure offering based on its UCS compute platform with virtualization support from VMware. Its management layer is based on software it acquired from its acquisition of Springpath. Currently, it is too early to tell how HyperFlex will measure up in the marketplace. Cisco may yet deliver on its promise to bring the network more fully into the management layer. Unlike solutions that are aware of the storage layer but are unaware of the networking layer, Cisco is in the unique position of being able to bring a fully integrated solution to market. EMC/VMware EMC offers VxRack under the VCE label and EVO:RAIL under the VMware label. VxRack is a pre-engineered solution for scaling out a Vblock-like architecture in smaller increments. It can be manufactured with different software personas, depending on your workload requirements. EVO:RAIL appliances have all of the common HCI features built entirely on VMware software technologies, including vSphere, vCenter, Virtual SAN, Log Insight, and the EVO:RAIL Engine for management. The basic EVO:RAIL appliance can be configured to support up to 800 VMs. DataCore DataCore has been in business since 1998, well before the birth of CI or HCI. While not making a direct claim to be an HCI vendor, they are worth a mention. Their SANSymphony product brings the distributed storage virtualization part of HCI to any hardware platform. SANSymphony can bring the significant performance improvements and data resiliency of HCI to IT environments without requiring the purchase of specialized appliances. It is compatible with VMware, Hyper-V, XenServer, and KVM.

Analysis At the present time, we believe there is no need look beyond this list. The other players in the market are either too young to have a proven track record or too niche to consider for general workloads. Nutanix is the leader in the space. Once Cisco’s HyperFlex matures a little more, it will be a serious player. SimpliVity is solid, especially if the uncertainty about UCS is not relevant to a business. DataCore is worth a look if all you need is the performance and resiliency gains of storage virtualization. EMC will appeal to existing EMC shops, though cost may be a factor.

DSScorp.com

7

DESIGN | HOST | MANAGE

Chapter 4: How Will HCI Impact My Infrastructure?

HCI can be implemented alongside your current infrastructure. And while there is no need to rip-and-replace, you may find it advantageous to migrate completely to an HCI design at some point.

Storage

Networking

Hardware Refresh Cycles

The most obvious difference between traditional storage and storage in HCI is the scalability of storage. The ability of some vendors to support adding storage-only nodes makes expansion a quick process.

Because HCI places such a large emphasis on distributed storage, your existing network may not have sufficient bandwidth to support it adequately. You should be planning for a 10Gb or higher network when implementing HCI. To control costs, you could build out a 10Gb network just to support storage traffic between your HCI appliances, rather than retrofit your entire network.

When working with traditional infrastructure, hardware lifecycle management can become complicated because of all of the variety and the sheer number of components. Since HCI drastically reduces both the number and variety of components, hardware refresh cycles become much

Along with improved resiliency and performance, HCI provides data protection across multiple drives and multiple nodes. A failure on one drive or node can be recovered on another. HCI also offers lower latency because of the use of SSD storage and the distribution of input/output operations across multiple nodes and disks in the environment.

DSScorp.com

8

DESIGN | HOST | MANAGE

Chapter 5: Ways to Apply HCI

Beyond transforming your IT infrastructure for better performance and resource optimization, HCI provides a stepping stone for implementing other types of technology. Below are some of the most common ways HCI can expand your capabilities. On-Premises Cloud The pressure within the IT organization to provide the speed of deployment and scale of the public cloud pair well with the benefits inherent in HCI offerings. HCI lends itself well to providing public cloud-like infrastructure in-house. Virtual Desktop Infrastructure (VDI) HCI meets the performance and user experience requirements of VDI users. In fact, some HCI vendors actually started in the VDI space. VDI users require fast response times that necessitate rapid input/output rates and low latency. HCI can deliver this at scale in a simplified and cost-controlled manner. Testing and Development HCI enables you to build out test/dev environments on-premises instead of in the public cloud. Budget constraints sometimes require that internal test/dev environments be constructed with inferior hardware or insufficient resources. HCI is cost-effective enough to allow for test/dev to be provisioned properly on premise. Data Management and Disaster Recovery HCI solutions often incorporate data protection features in addition to data reduction technologies. Highly effective business continuity solutions can be constructed without excessive cost and complication.

DSScorp.com

9

DESIGN | HOST | MANAGE

Conclusion: How DSS Adds Value As an emerging and complex technology, hyperconverged infrastructure can seem a bit daunting. But it is worth your time to consider how it might improve the capabilities and lower the costs of your IT environment. With HCI, you can boost the performance of your data center while consolidating your overall infrastructure. Because the journey to HCI will be overwhelming, you will want to work with a partner who specializes in IT infrastructure, someone who can help you navigate through the hype and help you make good, well-informed decisions for your business.

DSS specializes in IT infrastructure, and we are here to help you explore HCI technology and advise you on effective solutions to fit your needs. As a value-added reseller, we go beyond implementation to provide management and maintenance for your HCI. DSS is also in the unique position of being able to host all or part of your HCI solution in our data centers. Unlike our competitors, we can take you through your entire hyperconverged journey from consultation to implementation and ongoing management. See Where Hyperconverged Takes You Not only can hyperconverged technology transform your IT environment, but it can also function as a springboard for further IT initiatives. Why not rethink how your infrastructure can be opened up to enable new possibilities for your business?

Explore the potential of hyperconverged infrastructure

Schedule your infrastructure consultation today

10