1 When Virtualization Meets Infrastructure This paper explores how the rapid adoption of virtualization and cloud technologies are reshaping how we think about IT - and how it s transforming businesses today.
2 Table of Contents Executive Summary...1 Building Cloud Solutions...2 What s Driving the Need for Cloud and Virtualization Technologies?...3 Making the Most of Virtualization + Infrastructure, Layer by Layer... 6 Conclusion...8 Recommended Next Steps...9 References...9 2
3 Executive Summary Adoption of virtualization technologies in data centers is inching ever closer to 100 percent. No wonder: organizations value all of the -bilities that virtualization provides, such as flexibility, manageability, scalability, etc. However, for all the years that virtualization has been around, it has typically lived on premises: in the servers in your own data center. That is changing. Virtualization is moving to the cloud, and the combination of virtual environments both in the cloud and on premises is delivering new levels of service that enable IT to be more responsive to both users and business owners. Companies are increasingly taking this hybrid approach with virtualization to better manage technology spending while ensuring their data always remains available when their users, customers and business partners need it. This paper is for IT staffs who are under enormous pressure to manage their IT environments in the face of massive data growth, increased security threats and compliance concerns, and tighter budgets, all without sacrificing performance. That is a tall order, but it s one virtualization in the cloud can rise to meet. Here, we will detail what cloud infrastructure solutions look like, the benefits they generate, and use-case scenarios that illustrate what we mean. In the end, you will have a clearer idea of ways you can make your IT department better equipped to make positive contributions to your business. 1
4 Building Cloud Solutions With multiple forms of cloud computing available, each serving different purposes (see Figure 1), it s easier than ever for businesses of all sizes to address their unique needs. These cloud iterations include: SaaS Software as a Service PaaS Platform as a Service IaaS Infrastructure as a Service Most organizations require a hybrid blend of these to meet their business objectives. However, finding the right mix and balance can create a serious challenge, especially if your provider does not offer all three. Figure 1. Cloud Computing Iterations One size does not fit all: your needs should drive your solution. Most cloud hosting services respond to the market s desire to save money and/or increase production, but they don t necessarily address the overall completeness of a customer s vision and requirements. For example, how can cloud services create new revenue streams or decrease product time-to-market? Indeed, not all virtualization infrastructures offer a full suite of services, or the ability for integrated solutions within the direct offering, and not all problems can be solved through a single product. It s helpful to think about the cloud as a tool in the solution toolbox, rather than an end-point itself. Clouds however they re shaped are deployed in order to drive specific outcomes. We ll look at those drivers next. 2
5 Use Case: Health Care Data A health care provider ran into Big Data troubles when switching to electronic medical records (EMR). They invested in a system to support their existing digital data footprint and existing backup, only to realize after having spent a great deal of money that they underestimated their capacity requirements after their budget cycle closed. Naturally, this can cause major headaches. What s Driving the Need for Cloud and Virtualization Technologies? Driver 1: Data By 2020, the digital universe will contain as many digital bits as the physical universe contains stars. Doubling in size every two years, the data we create and copy annually will reach 44 trillion gigabytes. And by that point, the cloud will store or process 40% of that data. i The absolute enormity of those figures is hard to grasp. It s even more difficult to understand the burden it places on the modern business, which must extract the data that s actually useful (an estimated 22% to 35%) and then analyze it before they can realize any profit from it. ii Fortunately, cloud offerings can allow for spikes in demand or more flexible growth without necessarily requiring additional CapEx costs. Sources: Business Insider Intelligence, CustomerThink, IDC, Gartner, Wall Street research Figure 2. The Growth of Big Data In other words, as more and more data is being created and stored, businesses must adopt specific mechanisms to actually interact with the data and to have that data be useful; and they have to contend with exponential data growth year-over-year. Individually, the average successful business faces approximately 10% data growth each year (see Figure 2). Even worse, inefficient storage plans mean that data can be costing companies even as they fail to turn lead data into usable gold. This scenario can happen in industries such as health care where enormous amounts of data are generated frequently but not necessarily on a regular schedule. This makes it difficult to predict how much hardware you need. 3
6 Use Case: Legal Compliance A law firm needed both to keep case data more than 10 years and also have it be fully recoverable, with users able to search files quickly among the growing caseloads. They solved this problem by creating storage tiers that allowed them to keep hot data on premises and use the cloud to offload colder data that would be accessed less frequently. This balanced the need to be compliant against responsiveness to users needs. Driver 2: Value and Cost Savings Storage capacity costs are becoming trivial. Microsoft OneDrive for Business just increased its storage to infinity (unlimited) for subscribers. iii Amazon Web Services has instituted 45 price cuts in six years, and whereas a gigabyte of storage on a hard drive cost $9,000 in 1993, it cost just four cents last year. iv But if those numbers seem to remediate the implied costs of dealing with Big Data, they don t. The issues include both scale (pennies seem innocuous, until you multiply them by gigabytes, terabytes, exabytes and zettabytes) and use (it may be relativity cheap to store data; it can be cost-prohibitive to analyze and use it). Further, many companies are realizing that with compliance comes added complexity plus additional storage requirements. Sources: Business Insider Intelligence, CustomerThink, IDC, Gartner, Wall Street research Figure 3. Cost Savings Comparison Raw storage space is just the tip of the cloudberg. The average business keeps 85% of its data unused in what TechRepublic terms a data dungeon. v The cost of raw storage space may be pennies on the dollar but infrastructure costs combine both hard expenses (power and cooling alone represent one of the biggest line items on any IT department budget) with soft costs (like inefficiency due to server sprawl, when each server is capable of only doing one thing). For an illustrated breakdown of how quickly costs can add up, see Figure 3. The difference between the top and bottom halves of the chart is virtualization: by consolidating, costs fall dramatically. 4
7 Use Case: A Study in Maximizing Value A large university had just received funding for new storage systems and had them deployed locally. They soon discovered usage peaks and valleys, such as spikes during student registration and lulls during breaks. The peaks put significant strains on the systems. For them, the ideal solution turned out to be a hybrid cloud solution leveraging cloud and colocation space within their provider s datacenter. This gave them much better resiliency and put less strain on their local team for managing the systems. They were able to augment their physical deployment with cloud assets that were used to offload their fixed assets during peak periods. This also allowed them to have a disaster recovery sandbox to test and identify potential availability issues before they become a problem. CapEx vs. OpEx Cloud infrastructure allows your business to move to a fixed-cost model without having to worry about large fluctuations. Under a capital expenditure model, businesses face a huge uptick in spend every three to five years, when it s time to swap out all those aging servers, and then amortize the expense over time. CapEx plays havoc with cash flow; and in the IT world, purchases today must still be delivering value three to five years from now. Are you sure you know how much infrastructure you need to buy today to remain competitive years from now? Operational expenses, by contrast, do not require time to derive value. The purchase is expensed then and there. That kind of fixed-cost regularity is appealing to businesses, and virtualization makes powerful computing available on an OpEx basis. Not only that, virtualization can also flexibly integrate existing or newly purchased infrastructure and wring extra value out of them. Virtualization shines in driving value This scenario driving value and cost savings is where virtualization truly shines: keep what you have and augment it with virtualization to whatever degree required, because virtualization technologies scale virtually effortlessly. Of course, this only works if your cloud provider can provide flexible, customizable solutions. With a single-point solution, you re always either going to miss performance, functionality or equipment you need (subpar value) or you re going to have too much (waste). Unfortunately, virtualization does come with a caveat: if you consolidate your servers, you make yourself more dependent on fewer (i.e., if a single server fails, it would be proportionally more damaging to the business). That kind of exposure necessitates a solid disaster recovery/business continuity (DR/BC) plan with multiple layers of planning. We ll explore that next. 5
8 Making the Most of Virtualization + Infrastructure, Layer by Layer We ve already mentioned that not all virtualization infrastructure providers offer a full suite of services, such as disaster recovery/business continuity (DR/ BC) functions. While cloud computing platforms offer flexibility and a range of options, there is not a one size fits all solution for each objective, and fully leveraging the power of the cloud and virtualization requires combining the pieces into a cohesive architecture, with each component working in concert. Figure 4. The Virtual Layer Let s use a DR/BC plan as an example. First, let s divide the project needs into logical layers: virtual, physical and recovery. Layer 1: Virtual The Virtual Layer refers to the cloud you will use to accomplish the compute, storage and networking needs of your hosted applications and resources. This can be done via the Public Cloud (like Amazon), a Multi-Tenant Cloud through a service provider, your own Private Cloud (in-house or through a service provider), or a hybrid that exists both physically and virtually (see Figure 4). The Virtual Layer is the first building block of any cloud solution, answering foundational questions like: What kind of cloud is it? What parts are being managed by whom? How scalable is it; how is it going to grow? Layer 2: Physical The Physical Layer addresses where your company s cloud will physically reside an internal or an external data center? It s important to address the scale of physical architecture needed not just today, but also into the future. It s equally critical to pinpoint geography: just because it s in the cloud does not mean that location is unimportant! For one thing, availability and uptime can hinge on physical location: you still have to get to the cloud from wherever you re physically based; and if it s far away, you re likely to face latency issues resulting in a subpar user experience. For another, geography can expose the data center to additional environmental risks, like hurricanes or earthquake. 6
9 Use Case: Surviving Sandy A global enterprise working across all verticals built an impressive state-of-the-art data center facility. They did it right and passed all audits, even managing offsite backup. Then Hurricane Sandy arrived, knocking power out in eight states, affecting 7.4 million customers, and wreaking an estimated $65B in economic havoc. They even had a generator and extra machines, but they were ruined by Sandy (see Figure 6). The combination of cloud and virtualization technologies allowed them to restore their entire operation in hours rather than days. Layer 3: Recovery Finally, while the Virtual and Physical layers will backup and restore your systems, where will you recover to? The oft-overlooked Recovery Layer identifies the hot site for restoration/replication and includes non-cloud considerations like workstation and other resource requirements for your staff. Oftentimes businesses do not invest in disaster recovery because they re so heavily focused on production, production, production; which actually makes this another area where the cloud-plus-virtualization really shines. Virtualization By using virtual Disaster Recovery frameworks (e.g., a VMware virtual machine utilizing Site Recovery Manager), you no longer have to worry about procuring like-to-like hardware assets (i.e., two of everything), and you can focus on performance metrics like memory, compute, networking preconfigured whatever you need to run applications, not specific hardware. 1 Infrastructure Then, you can extend that value and ROI through to your infrastructure by addressing the costs of building your own cloud versus more efficiently buying space in a provider s cloud, and incorporating the efficiencies generated by virtualization. Together, it s possible to eliminate large capital spends and garner a fixed-cost model in the process, as previously discussed. In terms of DR/BC, we can also address the most often overlooked elements of a comprehensive DR plan. For example, how will you guarantee power when local power goes out? 2 Multi-site capabilities also enter the picture. With the right colocation partner, your data can live distributed across multiple physical locations, giving you that much more added resiliency. Figure 6. Sandy takes her toll. 1 If you re only looking at single-point solution, you have to make sure your production platform matches up with solution. For example, the provider may only be able to offer file-level backup, so you d have a solution, but it might not fit your operations. Then also, if you look at individual workloads and applications, they may not guarantee restore times that align with your needs. For example, though might take less than 15 minutes to restore, file services could be 12 hours. To meet specific Recovery Point Objectives and Recovery Time Objectives, you have to find a flexible provider. 2 Other overlooked elements include issues like compliancy when relying on failover equipment, networks and protocols; as well as data access when the workforce gets forced out of its primary environment. We should not forget the human component as businesses often do. Many cloud providers can also provide physical seating at or near their data centers. 7
10 Recovery in 4 Hours Instead of 40? Backing up your data and operations is one thing; restoring them is another entirely. In the aftermath of a disaster, even your spares may be unusable (aged or also damaged). And then, when you procure new hardware, you have to install the OS from scratch, configure it, install all the patches, all the updates, and only then can you install the backup agent. Only at that point can you start recovery. Altogether, you re looking at about 40 hours, per server. A virtual environment can accomplish the same in a fraction of the time (see Figure 5 illustration). Figure 5. Physical vs. Virtual Disaster Recovery Conclusion Virtualization and cloud technologies are reshaping how we think about IT. Gone are the days when IT was simply a cost of doing business. Today, virtualization and cloud are enabling organizations to be more agile and responsive to their customer needs. IT becomes less of a burden and more of an innovation platform. Challenges remain though. Cloud does not mean the same thing to everyone. There are different ways of implementing cloud solutions. It is becoming less and less practical to manage all of your own data; it creates too much cost variability, introduces too much risk, and puts a substantial strain on human resources. So what do you do? All-in-one cloud may not be for everyone, but you need to find the right balance for your situation. 8
11 Recommended Next Steps: Self-Assess. Reevaluate how you re doing business, and identify efforts that are no longer required versus business operations that are supercritical. Focus. Do you really need a deep IT bench provisioning servers, or would they be more valuable focused on your users, applications and business growth? Shift the daily care and feeding of servers to a provider. Ensure user experience. Research where your users fall in proximity to the data center hosting your cloud; similarly, investigate specific application workloads to understand where virtualization will actually help. Ask. Customers should expect a cloud provider will walk them step-by-step through their needs and figure out the right, tailored solution. If they won t, find another cloud provider who will. References i. IDC. (2014, April). The Digital Universe of Opportunities: Rich Data and the Increasing Value of the Internet of Things. International Data Corporation. digital-universe/2014iview/executive-summary.htm. ii. iii. iv. Ibid. Office 365 Team. (2014, October 27). OneDrive now with unlimited storage for Office365 subscribers. Office Blogs. Bort, J. (2014, November 9). Something called The Race to Zero is scaring a lot oftech companies. Business Insider v. Shacklett, M. (2014, August 25). Cold storage meets big data: run low-cost analytics in the private cloud. TechRepublic. data-run-low-cost-analytics-in-the-public-cloud/. 9
12 About TierPoint TierPoint is a leading national provider of cloud, colocation, managed services and disaster receovery solutions designed to help organizations improve business performance and manage risk. With corporate headquarters in St. Louis, Mo., TierPoint operates 13 highly-redundant, Tier III plus data centers in the states of Washington, Texas, Oklahoma, Pennsylvania, Maryland, New York, Massachusetts and Connecticut. To learn how TierPoint can help with your cloud, colocation, managed services and disaster recovery initiatives - call TIER (8437), or visit us at. TierPoint 520 Maryville Centre Dr. St. Louis, MO TierPoint, LLC. All Rights Reserved.