Cloud Migration: The Concerns and How We Should Approach It
Migrating your applications to the cloud has become a very clichéd scenario. It has been years since this strategy took its baby steps, and now it has grown and matured to the extent that most enterprises have at least one application or service running in the cloud.
In most cases in recent years, when you migrate to the cloud, you shift your entire in-house infrastructure to the cloud, where the vendor would manage the computing environment. However, this was quite a challenging operation for many organizations because they had to give up a certain level of control to the cloud vendor. And the move in itself was a challenge because there are certain factors in it that you have to be aware of. We will be discussing about that in the next section. First, let’s look at the most recent trends in cloud computing.
The major trends to watch out this year are:
Thanks to the adoption of Kubernetes and serverless computing, enterprise applications are deployed in a different manner. Kubernetes is a portable and extensible open source platform for managing containerized workloads and applications where automated deployment and scaling is done.
In 2019, many vendors are planning to roll out new features for API driven security policy orchestration to help deploy and scale applications in a secure manner. The modern core-business apps will base themselves on the foundation built through serverless computing and Kubernetes.
Serverless computing deployment will also gain traction as more companies look to roll out digital services and applications. In serverless computing, customers develop, run, and manage applications without building and maintaining the infrastructure to run them.
Private cloud is where enterprises host workloads that cannot be hosted in public clouds when issues of compliance, architecture design, cost and security arise. This year, three different approaches are going to be popular for private cloud hosting.
(1) DIY (Do it yourself) approach with VMware vSphere and software-defined infrastructure; but it is believed to be expensive and complex.
(2) Also, another DIY model where you use OpenStack open source software.
(3) You can have the cloud vendor build it for you with the help of converged or hyper converged software stacks.
Rather than depending on a single cloud platform to transfer all your data and applications, you can choose multiple cloud vendors. For example, companies like GE uses both Azure and AWS. This way the players can focus on their unique and specific offerings, because customers come to them for a particular reason, and they need to make upgrades in that area.
Industry-specific SaaS systems will be in the rise in 2019. This is because the SaaS ecosystem is just perfect, as the services-based architecture and the one-version approach is enabling enterprises to facilitate many-to-many customer scenarios and deep interactions. This aids in intracompany collaborations and the ability for multiple companies to connect in real time. Companies like AthenaHealth and Salesforce Financial Services have already done it.
While migrating to the cloud does have its benefits, there are still certain things to be aware of if you want to make the transition smooth and hassle-free. There are challenges that you might meet every step of the way, and it is a good idea to be prepared so you won’t be caught in any nasty surprises along the way.
Here are some of them:
Moving to the cloud is a huge business strategy that you cannot take lightly. Improper planning or not recognizing where you might go wrong can spell doom. To adopt the cloud and implement it correctly, you need to have fool-proof planning, because there is no generic solution. Sometimes it becomes difficult to migrate the applications and data, and you may have to start again.
How to Solve: Go through your enterprise’s current infrastructure minutely, and see what needs to be integrated and how. Certain adjustments and rebuilding may have to done for the applications to work once they are in the cloud. Divide the migration strategy into small bite-sized chunks, and you can make your cloud journey smooth.
Vendor lock-in is a proprietary lock-in situation where the customer dependent on a particular cloud service provider (CSP) for certain products and services is unable to switch to a different provider without huge switching costs. Moving to just one cloud vendor can be a risky business move because the vendor can experience outages, and that might, in turn, cause serious business repercussions.
It is all your money, infrastructure, data and applications that’s at stake. It is not that outages happen often, or there would a problem when you try to transition your products and services, but it is better to be prepared.
How to Solve: There are a few ways in which you can combat this problem. Here they are:
(1) Compare the different cloud vendors based on their service offerings, negotiate on the entry and exit strategy and see whether would assist you with reconversion when you want to move to another vendor.
(2) If you are not particularly happy with the vendor, then make sure you don’t auto-renew their contracts, because vendors usually auto-renew the contract once the period is over.
(3) Keep a backup vendor or have a secondary relationship with a different vendor, so if you wish to leave the primary CSP, you can always have the other option.
(4) If you have a PaaS vendor, then you can easily decouple your application from the vendor’s underlying infrastructure. This can be easily done if you design portable applications.
Security is the biggest worry for most CIOs when they shift their data to the cloud. Enterprises often get cold feet when they consider moving their confidential data to the third-party provider. Even when they hear that the cloud environment is secure, they still have doubts. This could be a problem, especially since you know you have to move to the cloud.
How to Solve: CSPs have advanced systems these days. For example, AWS spends time researching on trying to track and fix known security hacks with their APIs. They have highly secure data centers with strong safeguards. And Azure too has a security detail that’s much more than physical controls or cybersecurity experts. Microsoft invests a billion dollars for cybersecurity, and they build customized hardware with updated security controls within the hardware and software components.
Other CSPs have their own technology and security measures to safeguard your resources, detect threats and prevent attacks. Make sure you discuss this in detail with them, before moving.
This is a fear most enterprises have. Failing to analyze cloud computing strategy can lead to exponential costs, and at the end of the migration you will be presented with a huge bill, not to speak of the monthly bill needed to maintain the infrastructure.
How to Solve: You can save a lot of money by evaluating the cost of migration and operation or rather the TCO (Total Cost of Ownership). The cloud vendor will charge you for network traffic, storage capacity and computing. Sometimes, they also provide additional costs for certain services, or for not using it.
For example, AWS provides Elastic IP Addresses (EIP) that is actually a permanent IP address for an application. The funny thing is that when you are using the EIP, you don’t have to pay for it, but when you are not using it, you have to.
You can also save money by assigning the tasks to different cloud providers depending on the services they provide.
It is also advisable to assess the TCO under different load patterns. Depending on the season, your application is likely to experience load variations, and more often TCOs will be based on static pattern of use.
Apart from the above, ROI (Return on Investment) should be calculated.
Poor strategic and architectural choices can cut down all the cost advantages. Most people think that the "lift and shift" cloud transition is perfect for them, but it doesn’t suit everyone. In the lift and shift method, you upload virtualized images of the present in-house systems onto the cloud service provider’s infrastructure.
How to Solve: It is costly and ineffective when you place an entire enterprise architecture directly onto a CSP. You might have to invest time and effort to design the enterprise architecture, but it would be definitely worth it.
Uptime can vanish in the cloud as well. Amazon S3 went through a major outage after a small command error was made. It did not just affect the storage service, but other Amazon services as well, and even for providers who built their services with Amazon as base. What will you do in such a circumstance?
How to Solve: Vendors may claim that they rarely have an outage or boast of better uptime statistics, but make sure you draw clear service-level agreement and understand the risks involved before going forward.
After analyzing the risks and the possible ways to get around those challenges, you now have the real task ahead. And that is one of the most crucial part of cloud migration - Make sure your applications are ready for the cloud. Unless that part is totally cleared, all other efforts go down the drain. Here’s how you do that.
Complexity in the application’s design is a major consideration. It might not be easy to migrate traditional applications as they are usually complicated and tightly coupled. The customers might also not be willing to rework them. Henceforth, it is important to have a distributed architecture and scalable design to enable cloud migration. There are tools that would help you analyze the cloud-readiness of your application.
Applications have integration points like SMTP servers, web services, third party vendors, payment gateways and external storage. It is important to check these integration points and how the migration process would affect their dependencies. Once you pinpoint those areas, you can rectify the challenges that arise there and move forward. The challenges normally would be in the connectivity or authentication spots.
Understandably, database migration is no small task. The cloud provider must be fully equipped to deal with migrating petabytes of data. They should also have the provision for rollback in case of any unexpected challenges. Being a critical part of any application, it is important to check whether your cloud provider can do this without any glitches. Services like Azure, and even other third-party vendors do provide seamless database migration to their customers.
Multicasting is a method of sending information through a network. Unicasting and broadcasting are the other methods. When information is sent through multi-casting, a single transmission will be split up among different recipients. The sad part is that most of the public clouds do not support multicasting; so if that method of ‘information sending’ is important for you, you need to rethink using the cloud.
If your application runs on a different operating system from that of the CSP, then you need to check whether migration is possible. If compatibility between OSs is an issue, then you might have to find a suitable, compatible OS to run your applications.
Moving to the cloud is inevitable, and finding a fail-proof cloud migration strategy is definitely a strategic move for your business. You need to understand the risks and problems involved in the migration program, find out which CSPs are capable of providing what you need, assess the applications to see whether you can do it seamlessly and without compromising on the security, functionality, reliability and scalability of the application.
Still need clarity on how to proceed with cloud migration? We can assist you!
Related Articles
