Weighing a Cloud Migration? Get Answers from An Industry Veteran

As Director of Engineering and Operations at Green House Data, Mike Mazarakis has helped his share of companies migrate to the cloud. With 20 years of data center and networking experience, he's a self-described “pragmatist in IT” who has watched virtualization evolve into the concept of cloud we all know today.

Mike answered questions submitted by the public in a webcast last month. We interviewed him to get the answers to the most pressing cloud migration questions and help you plan your move to hosted IT. Look for more features in our cloud migration series in the coming weeks.

Tell us a bit about your background with virtualization and IaaS.

I have been working for organizations that used virtualization for about ten years, and I have been heavily focused on the server, storage, and orchestration side of virtualization and cloud for the past five years. My background covers enterprise, service provider, and e-commerce environments.

I spent 15 years as an Engineer with responsibilities including networking, security, system administration, and data center infrastructure. Most recently, I ran the Data Center and Server Technology Operations for RealNetworks’ gaming division, GameHouse, which spanned across the US and Europe. After that, I ran the Engineering and Operations teams for FiberCloud, a colocation, network, and IaaS/SaaS service provider based out of Seattle.

What is the most common cloud migration scenario that you see with customers who are new to the cloud? Do they start with one or two apps like e-mail and file storage, or place their entire IT in the cloud?

Most new cloud customers generally fit into three categories:

Startups or development teams that need to quickly deploy resources.
Generally, these customers will put everything into cloud or SaaS platforms, and try to run as little as possible internally.

Customers who are entering a technology refresh cycle for legacy hardware.
These customers typically have an application or platform that is approaching end-of-life or end-of-support, and they need to migrate or upgrade the system. In this case, they will outsource either the application or systems that are running on the legacy hardware.

Customers who are trying to stretch their IT dollars further.
They may have a reduction in staff, or their IT systems are growing quickly in size and scope but they lack the budget or inclination to add additional staff to support the infrastructure. They generally follow a similar model to technology refresh customers.

Are smaller or larger businesses more likely to go all in on IaaS?

Startups and smaller organizations are very likely to go all-in on IaaS and SaaS platforms. It generally doesn’t make sense to incur the capital costs of building out servers or an entire datacenter when you can tap into resources that scale as you grow.

Medium sized organizations can go in either direction. They generally hit a tipping point that pushes them either into investing more heavily in their existing infrastructure or outsourcing that infrastructure. It very much depends on the budget and resources of the company.

Most medium and large companies tend to look at IaaS on a per-project or per-platform basis. They also might consider SaaS or IaaS when they start a technology refresh cycle. They find that the TCO of refreshing their equipment, maintaining the associated infrastructure, and staffing far exceed the cost of outsourcing the system.

Walk us through a migration scenario where Company A is moving their on-premise infrastructure to the cloud, but they want to continue using some of their existing systems and IT assets.

Company A has a decided to upgrade their Exchange Infrastructure from Exchange 2003 to 2013. They currently support 20,000 users, and have found that their existing Servers and SAN infrastructure will not support Exchange 2013.

Company A then engages with their preferred hardware vendor and provides their specifications to purchase new servers. The vendor responds to the RFP with the standard reference architecture for Exchange 2013. The total cost to the company is $600,000 for the hardware, operating system licenses, and network upgrades. Annual support is $50,000 a year.

This is before you get into the cost of adding Exchange itself, anti-virus and anti-spam platforms, the costs of training your staff on the new systems and hardware, and the costs of performing the user migrations. Let’s assume that the combined cost of the infrastructure, software and licensing, training, migration, staffing, and support is $3.4M over 3 years. You are likely looking at $900,000 in staffing cost alone to maintain the environment. This may be an optimistic number, but we’ll go with it.

Company A then looks at two additional options: a SaaS-based platform and an IaaS platform. Pricing for the SaaS platform comes back at $5 per user per month, or $100,000/month. For an IaaS Private Cloud Solution, they might be looking at $85,000 a month for a dedicated environment.

The SaaS platform likely offers you a way to start migrating users almost immediately, but for an additional cost. The IaaS platform gives you more control based on your security policies and costs less, but may have a three month lead-time to deployment. In either scenario, you are likely saving money and increasing reliability by outsourcing this platform instead of running it yourself.

Company A chooses IaaS, since it offers greater cost savings and allows them to extend their security policies to the new environment. They coordinate the implementation of the new hardware and software with their vendor. Their IaaS vendor has on-staff talent with Exchange, so they’re able to deploy it quickly and at less cost to Customer A. They provide Company A with a migration plan and tools to move their users, which Company A’s existing IT staff performs.

After four months, all of the users have been migrated to the new platform, and Company A is now able to focus their staff on the continued operation of their other infrastructure, such as desktop systems, Active Directory, and Sharepoint.

When starting a cloud migration, what are the most important assessment and planning considerations?

The following are questions I typically ask customers looking to migrate to the cloud:

• Are you migrating an existing application or building a new application?
• How many users will be accessing the application(s) and via what method(s) (VPN, MPLS, Direct Internet Connection). This will generally drive the rest of the requirements based on the application vendor’s specifications.
• Have you created a list of all the systems that need to be migrated?
• How many virtual machines do you need? Many smaller applications can get by with a single VM. Larger applications typically require several VMs to split out their various functions.
• For each server, how many CPU cores do you require? How much RAM? How much disk? Do you have a specific IOPS requirement for systems that run databases or reporting?
• What the security requirements around your application? Firewall, data encryption, auditing, etc.

What are some of the technical aspects of moving from on-premise to cloud infrastructure? Setting up IPs, port forwarding, attaching storage…it can’t be as simple as copying over a virtual machine from an already virtualized environment.

• What are your firewall and security policies that need to be implemented?
• Do you know the firewall platform that is provided by your cloud service provider? Is it a full-featured firewall, or a simple one that is interfaced via API?
• How do you plan on migrating data from your premise to the cloud? Do you have enough bandwidth on your existing Internet connections to be able to transfer the data in a reasonable amount of time? Tools like DoubleTake from Vision Solutions can mirror a local server to a cloud server.
• Do you need synchronous or asynchronous replication of your data to the new servers?
• Are you exporting existing servers or creating new ones? If you are exporting, what format does your IaaS vendor use for importing your Virtual Machines?
• Will you configure a VPN to your IaaS vendor? Or access the platform via HTTPS/TLS?
• Do you have compliance requirements for PCI, SOX, HIPAA, or SSAE16? If so, do you have a plan in place on how to audit your vendors?
• How do you plan to monitor and alert on the systems?

Are there any scenarios where you would advise a company NOT to migrate to cloud infrastructure?

If you have a number of existing data centers with reliable and up-to-date infrastructure, along with the staff available to support them, you are generally better off continuing to maintain those systems. The tipping point of cloud vs on-premise systems is usually 100-200 virtual machines. Once you exceed that number, you may be able to save money by running the applications on your own infrastructure.

Is the cloud inherently less secure?

Not if security is properly implemented in your design. Almost all cloud providers support some type of firewall, VPN, and encryption. Backups and Business Continuity/Disaster Recovery are still an important aspect to consider as part of a cloud migration as well.

Anything else you want to add about cloud migrations?

Remember that the cloud is a tool, like any of the other items used in IT. Don’t look at the cloud as the enemy or a cure-all for your IT woes. Regardless of whether you use an internal or external platform, you still need to implement proper planning and administration of the infrastructure!