Hristomir Hristov: Cloud Migration – an Obscure Journey on Its Own or a Well-Paved Road

Hello. My name is Christ Christo, and I'm senior solution architect at co builder international. I will tell you about the experience gained to a couple of companies to acquire migration, whether it's an obscure journey on its own, or it could be a well paved road to a cloud transformation. My first qu migration included a couple, a bunch of virtual machines with web applications, big volume of documents, some relational databases in a routine based search, and it should be moved to Azure cloud back then, eventually it appeared in years.

Then the public cloud provider was not providing, was not supporting sticky session out of the box. Put another words. If a user walks into one note, it is not sure that all the subsequent requests will go to the very same note. And this was a broker for applications relying on stickiness of the session. So the public cloud provider was giving an official solution. They called it an official hack, and we ended up in Microsoft's office in Munich for one week workshop, but let's return nowadays and see how the Corona crisis have affected the cloud transformation.

In the days when the consumers are forced to stay at home and he, with our work from home policies, it, it appears that the COVID 19 crisis have significantly increased qu adoption. Actually more than half of the respondents say they have in significantly increased their plans to move to the qu rather than they expect it in the beginning of year. And according to Flexera, state of the qu report for the previous in this year, it seems that enterprises are embracing the multicloud strategy. Totally. It's about 93% of the respondents say they see this the right approach.

And especially the hybrid one where at least one public and one private quote is used again to this report. It seems that almost organization used at least one public or one private quote it's 19% and more over the newcomers who, who joined the qu transformation be on the hybrid approach since it gives both best of breed features between the providers and avoid the risk of walking to one vendor.

However, organizations challenge challenges are increasing in the, this new multi hybrid quality environment and seamless governance and management of the tools is something that should addressed accordingly. Currently, just one third of the organizations use multicloud management tools like Amazon web services, outpost, Google on tos.

However, this number will significantly grow and in it's not surprisingly that in the last few years, public big, big public providers have provided such such control plan such tools in order to both manage and govern the resources within the primary quote and also in the rival quotes. So it's something that enables the company both to govern in multi as well as OnPrem environment, and also a single security posture is easily achieved. A successful call journey depends on assessment migration and operation.

I will not focus on operation based in my talk where you, you have the cost optimizations, the compliance, the continuous management. And we'll talk about the assessment in the migration. The assessment is the foundations of qu transformation. It is the very beginning where you assess whether the organization have reached quote readiness. And in order to assess this, a migration plan should be prepared, which should be presented to the stakeholders at an early stage, and also keep clear visibility over the costs in time.

Immigration point on its own could be really huge topic with a lot of details and give some highlights and important crucial aspects of it between one quarter. And one third of the servers in the data center, zombie servers, according to Gartner research, it means that those servers are not used at all. Moreover 40% of the racks are under provisioned and servers are operating not to their optimal capacity. At most of the cases in addition, hardware is bought for 2, 3, 4 years ahead of scaling up and down and to these factors move the companies to a cloud transformation.

There's another important moment of network and infrastructure architecture. It may seem alluring to someone just to move things as it is and copy the infrastructure.

However, it's advisable to architect in order to be cloud native infrastructure architecture here is show the classical approach of defending the Caso with the subnets network, security groups, virtual machines, and all balancers in front of them. This model assumes that the good answer inside the security parameter and the better answer outside, however, in the world of internet of things, mobile application applications, multicloud, this security paradigm is not one it's almost vanished, it's exported.

And, and nowadays it's all about the context between users data and the resources it's defining this context and verifying and continuously enforcing this context in order to resolve incidents and izing and improve the process, working in an environment as it is nowadays from home with users, machines, wifis, remote taxes, and the multi and hybrid quality environment as well, increases the complexity and shifts away the, the security from networks to focus more on other areas. For example, it's more about the data force now than the networks.

And it's about whether this is the real identity that should access the, the resources and the assets rather than how via what channel are our accessed all assets. And it's, it's all about the zero trust and the context that is continuously reinforced, including stakeholders at an early stage in keeping them important is crucial in order to go into one direction and have a smooth process to the quote. According to KPMG research, the biggest quote migration obstacles are politics, lack of alignment, lack of strategy and vision, inability to act on business critical signals.

And if it was five or 10 years ago, the lack of experienced employees or the maturity of the features, the cloud vendors nowadays, it's more about the Whyman the strategy and, and how we act on critical signals.

For example, the Corona crisis is such a critical signal to, in all aspects of our wife, including the digital and the business wife, of course, it's about budget and most big public providers provide tools to calculate the total cost of ownership, but there are four general steps to audit your current costs to calculate the estimated costs to the, to estimate the migration costs and to consider the post migration costs. And then it's the migration. The migration itself could be achieved in different ways.

Here are three of the most commonly used migration paths re hosting platforming and architecture. Rehosting costs are known as lift and shift is just to change the hardware environment without changing the applications, infrastructure, it's fast, less resource intensive process, but it does not benefit from quote native features. And it's more costly in time rather than other migration paths.

It's just, the infrastructure is a service way. Here's some examples, a web application could be requested on a virtual machine or availability set as infrastructure is a service or SQL server could be re hosted to a virtual machine or as a platform, as a service Azure SQL service managed instance. I'll give my example within the terminology of Azure cloud, but it could be Amazon web services or Google cloud platform here report for, from the other hand is just to repackage the assets and move them to the cloud. Usually it requires more DevOps work rather than development work and code changes.

It's a more feasible way to migrate to the quote rather than arching the resources. And usually companies find it viable. And one third of assets are moved. That way you can benefit from native features such as how to scaling availability containers, when report for me, however, there's some transitive risk. And also it could be a framework. Walking also features not should be taken into consideration. And here is an example from my first quote migration, the Luine based search was saw virtual machine, which was moved to the quote on a highly available cluster virtual machines.

The cluster was synchronized between each other of the notes by zookeeper and the wall bouncer was sitting in front of the quest in order to distribute the wall. So the search itself was not changed, but it resulted in a highly available service with zero down timeline upgrading, and it was only DevOps work. Another examples here could be web application, a sub-service or SQL server Azure SQL server. And we come to the rare architecture approach. It is also known as the refactor refactoring.

Usually the applications and resources are divided into parts in order to be more manageable and just one or certain parts are architecture and made to quote native architecture. It benefits from the quote native features.

And in, in that way, you can gain productivity. However, this is a tailored approach and it may result in a framework walking to a specific service, for example, Azure Kubernetes service or Google cloud Kubernetes engine, but it's more cost effective than the previous waste of migration. An example here could be an entire web application with a database and just the business logic to be refactor test microservices, or the front 10 layer could be refactor test single page application. There are also other ways to migrate assets to the quote.

You can revise them and just do small modernization changes in order to gain time and consider, which are the other approaches is best. How there are some upfront development work and also the legacy should be supported.

The, it it's continued to be supported if we have two web applications and to databases and a schedule job is synchronizing the data between the two databases. This schedule job could be replaced by enterprise service bus in order to achieve real time and data freshness. And of course you continue with your legacy assets. You can always and start from scratch, throw the quote out and make a quote native architecture. It's a way that gives you the opportunity to benefit from all native features.

However, there's again, possibility to framework walk in. And also you may find yourself in a situation that you have abundant, abundant cure assets and work. If there are some major problems through why rebuilding, if we have for web application data, ports documents, and again, scheduled service moves them to a service to be processed and to be ready for a second web application. This could be reviewed by quote native features such as rest based storage and broker messaging. And when a document is supported.