Lowering operational costs has always been a business goal for all enterprises, and migrating into the cloud delivers on that promise every time.
Additional perks of cloud infrastructure are lower risks, infinite scalability, high availability and better security. Despite these well-known advantages, some enterprises still hesitate with the move, afraid it will prove expensive, risky and time-consuming.
The Main Stages of Cloud Migration
Each stage of cloud migration is a process that involves solving technical and non-technical issues, facing certain challenges, and avoiding obvious or hidden risks. Before you make the move, you must learn what aspects you should pay special attention to during each migration stage.
- Checking Inventory and Choosing the Cloud Model
Inventory. Taking a full inventory of the whole infrastructure (and not just applications) is very important. You have to know all application components, how they work and interact with each other, as well as organise documentation for all those aspects. If there are gaps in technical documents, those have to be filled at this stage. Relying on detailed, full and accurate documentation facilitates migration in general as well as testing of all systems in particular.
Model. Choosing the cloud model – public, private or hybrid – is also a vital step. The public cloud fully relies on the provider’s platform, the private cloud uses the client’s platform, and the hybrid cloud combines the two models in one seamless solution. Statistical figures show that small and medium businesses can reap most benefits from the hybrid model, but a case-by-case decision should be made in each situation, opting for the model that is most convenient and helps reach the desired outcomes with the least amount of effort and expense.
Scaling into the cloud is the best migration technique, regardless of the cloud model chosen. Even when migrating into the public cloud, it is more reliable to scale your infrastructure into it, check if it functions correctly and then finish migrating the rest of it. Providing elasticity to your migration process will make it easier to scale the infrastructure, opening up stress-free opportunities for companies to purchase add-ons from providers when they need to expand a feature or application.
- Test Migrations and Cloud Assessment
Testing your migration and choosing a provider are very important phases for financial and legal reasons.
Test migrations. Taking baby steps and testing small amounts of resources to check if expectations can be met fully lowers your risks and helps you deal with small errors at the testing stage (where they can be fixed and avoided) as opposed to facing bigger problems further down the road.
You should test how scalability works in the cloud. Scalability can be performed horizontally and vertically. Generally, horizontal scaling means adding a feature to existing resources. For example, if a virtual machine is not powerful enough, another virtual machine is added for the second part of the application to achieve the necessary balance.
Horizontal scaling in the cloud is not as convenient as it may seem; it is better to use the benefits from providers that can guarantee more resources to a single virtual machine. But again, each migration case is unique and has to be assessed individually.
Provider assessment. Every provider offers a unique specter of services in addition to basic functionality, and you have to be sure you can reach all your business objectives with the chosen provider before you sign anything at all. How different providers guarantee security is a vital aspect to consider before making a final decision. For the hybrid cloud model in particular, the choice of the right provider is a critical, and often tricky, question to ask. In the hybrid model, you have to look for options to reserve channels and make sure that your dependence on network connections doesn’t impact functionality. Choose a private model if dependence upon the Internet is out of the question. Whatever you decide, this is a decision that should be made during the testing and assessment phase.
- Proper Planning and Finalised Roadmap
Taking into consideration the inventory map that was created at the first stage of the migration, a finalised roadmap for migration should then be created. The roadmap helps control all of the remaining stages and avoid chaotic, impulsive actions. It must include full details about all systems, providing obvious checkpoints for migration participants to check off one by one.
Choosing the right time for migration can often cement the success of the whole endeavor. With large infrastructures, there is never a good time to migrate, so the most critical objects and their work times must be outlined, and all other time slots should be considered migration time. Those less critical times can be sacrificed for a higher goal. Migrating to the cloud is always an exercise in assessing risks and benefits.
The best way to migrate is to scale into the cloud. If the client has its own server or data center, a full transfer can be too risky, given the possibility of a complete system failure. It is best to create a safety net – partitioning all systems by keeping certain aspects in-house and moving others into the private cloud. This process involves many risks, but the general recommendation when transferring a mail system to the cloud, for example, is to remember to keep network and storage architecture intact, and know how providers or clients will respond in the event something goes wrong.
However, some applications are impossible to scale up or portion out, and a full migration is the only option. The best tip for these cases is to never change anything during a migration. If one of the components has to be changed or updated, it’s best to do it after the successful migration. Doing so during a migration makes it impossible to detect what caused a malfunction – the move into the cloud or the changes and updates.
Security in the Cloud
The biggest problem in guaranteeing security is carrying out a serious audit of everything happening in the cloud, as well as system control. When I say “problem,” I mean that, while being one of the key steps in providing control and security that you should never skip, carrying out a comprehensive audit is a financial and technical problem.
To guarantee security, companies have to face the necessity of spending quite a bit on a quality audit and access control to the virtual environment. The best option is to choose providers that offer audit options and know in advance how those will be carried out. If that’s not a possibility, doing it on your own can prove to be expensive and difficult. It’s also paramount to pay attention to the security of network communications, the channel between your office and the cloud provider. Transferring data within the secure walls of the company is one thing; doing the same via a channel is another story altogether. Examine the channel for attack vectors and security loopholes to stay out of trouble.
Always check the authority of the cloud provider. Ask yourself the following questions:
- Do big clients trust this provider?
- Was it involved in any big scandals?
- Does it have enough experience to service my project?
Trust actions, not words.
The Main Factors that Guide Cloud Migration Strategy
Obviously, each migration is unique, but there are some factors that almost always influence migration to the cloud. Here are some of them:
- Scalability. If an app or infrastructure is dynamically growing and constantly changing, and it’s hard to predict how it will consume resources in the future, your migration strategy should safeguard the system during those fast spurs. If the infrastructure is static and the same amount of resources is spent without spikes and peaks, then you can do a full migration to the cloud in one step.
- The goals of infrastructure users. Users always have goals and expectations. In the case of a bank migrating to the cloud, for instance, a big part of planning should revolve around providing access to users. For uninterrupted uptime, you may need to make slight changes to your architecture or even fully redesign applications and infrastructure.
- Business goals. Considering these is an absolute must. Ask yourself questions such as:
- How expensive is down time?
- How important is constant availability?
If availability is very critical to the business, then it is better to stretch migration to six months, or as long as necessary. If eight hours of down time are not as vital when compared to the economic benefits of the cloud, then the eight hours can be sacrificed to move to the cloud and start making more profits right away.
How Not to Lose on Performance of Applications in the Cloud
Testing and studying providers should usually help you not to lose on performance. Choose a reliable provider with a trustworthy reserve channel, and realistically understand what services they can provide, at what times and in what amounts.
- Quantify your resources. The cloud helps you quantify your resources, where every resource can be carefully measured and you can easily see where every dollar is going. On paper, asking the provider for more resources for a short period of time may look innocent enough, but in reality, over-the-threshold consumption can cause severe financial penalties. If you predict peak times, make sure asking for more resources will not be so expensive as to make the whole process unprofitable. Ensure you know all the ins and outs of your agreement with the provider and what their plans entail (especially the small print).
- Sign service level agreements (SLAs). Good providers have solid SLAs that outline particular guarantees, which can help minimise spending on 24/7 in-house IT support. If there is a serious situation like hardware failure, that is no longer your problem. There may be discrepancies between what the providers may be willing to do and what they can do from the point of view of their billing policy. And then there are your interests. The common conflict arises when you want to pay a fixed bill at the end of a payment period and enjoy a guaranteed number of services while the provider wants to bill you for the resources you have actually used as part of a pay-as-you-go scheme. You should think about these conflicting interests during the stage of testing providers; lay conditions down in the SLA, together with possible back-out plans and alterations in case you need to scale infrastructure.
- Know and plan peak times. You should know and predict peak times on your end, as well as understand that providers also have peak times, which can impact system functioning. If both peak times coincide, you won’t have enough resources, and action should be taken to plan what must be done in such situations.
Costs of Moving To the Cloud: What Goes into the Equation?
- Downtime. Rewrite applications to back them up during downtime. If you have an infrastructure working with a private client software, moving it to the cloud without modifications is not possible, and applications should be either scaled, rewritten or otherwise adopted to work in the cloud seamlessly. For example, when moving part of a mail system infrastructure to the cloud, failures may occur if Internet channels are unreliable, which results in loss of client and staff letters without any retrieval possibilities; this is detrimental for business. Email architecture has to be redesigned in order to rent out part of a cloud email client infrastructure that guarantees high availability and no down time together with a trusted channel to which you are always connected. In case of interruptions, letters are stored in the cloud, and as soon as the system is back up, they are sent back to you. In a perfect world, provider networks should be partitioned to such a degree that lack of electricity at your central office should not influence user experience, and any user with an Internet connection should have access to the network.
- Electricity. For nations where electricity is not cheap, that’s a very important part of the equation. Moving to the cloud is extremely productive in these cases, because the money you spend on 10 physical servers can get you 20 or even 30 virtual ones.
- Reserve copies. Creating backups may be more expensive in the cloud. When the infrastructure is in the office, it’s easier to carry out a backup by adding a repository or library and copying data to it. In the cloud, it can be more costly, but don’t try to save money on this aspect since it’s always business critical. Rather, calculate your costs accurately and discuss this with your provider to negotiate a separate plan created for backups.
- Integration. If the cloud is public, you will run into common integration issues summarised by the following questions:
- How will users have access when it’s not on the local network but on the Internet?
- How will my team realise this?
- What technologies and network equipment should be used?
Pay attention to network equipment, changing network architecture by scaling it up or down, and making it more or less complex.
If you are working within the hybrid cloud model, then think about integration between the cloud parts of the infrastructure with the local parts. Ask yourself the following:
- How will the local and cloud infrastructures react to their roles of “pseudo neighbours?”
- Will all the applications work as intended?
- How will users react to the two infrastructures?
Also, remember that network issues may not be noticeable to the end users, but they do make a difference to you financially and are a very important part of the equation.
Clients who don’t move to the cloud always spend more than those who do. Just think about operation costs to service data centers, equipment costs (old or broken equipment has to be replaced constantly), and staff costs (both high and low level specialists have to be hired). Any time a large in-house infrastructure demands a complex redesign, you have to outsource specialists, which may be very expensive.
Even with the media frenzy over the recent cloud hacking scandals, migrating to the cloud is the way to go. In short, it’s the future of IT.
For those decision makers who still think of the cloud as something ephemeral and therefore risky, there is the hybrid method to provide extra peace of mind, allowing them to store sensitive data in-house and move the rest to the cloud. For every challenge the cloud may present, there is a working solution.
Maxim Sovetkin, System Engineer, Itransition (opens in new tab)
Image Credit: frank_peters/Shutterstock