dcsimg

Improving Your Cloud Deployment: 8 Key Cloud Rules

  • Improving Your Cloud Deployment: 8 Key Cloud Rules

    Clouds
    These best practices for cloud deployment can help organizations simplify and streamline the process of running production workloads on public cloud services.
  • 1. Re-architect when you migrate.

    architectural drawings

    It's certainly possible to take legacy apps and simply move them to a public cloud service (a process often called "lift-and-shift" by cloud computing experts), but this approach might not result in any net benefit for the company.

    Many cloud consultants say you should start the process by considering the reasons why you want to migrate your legacy application. Are you hoping to achieve greater scalability? Are you trying to control costs? Is the migration part of a larger digital transformation initiative or a push towards greater agility?

    If the answer to any of those questions is yes, organizations might be better served by re-architecting or re-working an application, either before you migrate or after you have "lifted and shifted" the application to the cloud. This re-architecting definitely increases the work involved in migration, but if enterprise plans to continue using the legacy application for a long time, the time spent in re-architecting will pay off in terms of greater scalability, flexibility and all the other benefits offered by the public cloud.

  • 2. Consider bare metal for legacy apps.

    metal gears

    Speaking of legacy app migration, you might want to consider a bare metal cloud service when moving existing applications to a public cloud. For those who don't know, a bare-metal server is a single-tenant server where virtual machines (VMs) are installed directly on the hardware (the bare metal) instead of in the operating system. (In others words, it's basically the kind of setup that IT departments used before virtualization and cloud computing became popular.)

    Bare-metal cloud services have some tradeoffs. On the one hand, they can offer faster performance than other cloud services for some workloads, and at least one vendor — IBM Cloud — will allow you to configure your bare-metal server to the exact specifications you need. That means you can exactly replicate your current data center in the cloud, which is helpful for lift-and-shift deployments. On the other hand, bare-metal isn't as scalable, flexible or inexpensive as traditional infrastructure as a service (IaaS). Still, it's worth investigating this option when migrating legacy applications.

    IBM and Oracle offer the best-known bare-metal cloud services. In addition, Amazon Web Services and Microsoft Azure have both announced previews of bare-metal services, and some smaller cloud providers also offer bare-metal options.

  • 3. Consider serverless for new apps.

    Computer screen with code

    If you are deploying a brand new, cloud-native app, experts say you should consider a different type of public cloud service that is more or less the opposite of bare-metal: serverless computing. The name is something of a misnomer because serverless cloud services still run on servers. However, the developers writing the application don't have to think about or configure the cloud instances they are using. They simply write the code, and the cloud service handles the rest, configuring and scaling cloud instances as necessary.

    The serverless option is becoming increasingly popular, especially with developers building applications with microservices architecture or developers who work in DevOps environments. The best-known serverless option is AWS Lambda; other options include Microsoft Azure Functions, IBM Cloud Functions and Google Cloud Functions. Also, Oracle has an open source serverless project called Fn.

  • 4. Containers make cloud deployment easier.

    shipping containers with cloudy sky

    Use of containerization tools like Docker and Kubernetes has skyrocketed recently. In the RightScale report, use of Docker increased 40 percent between 2017 and 2018, while Kubernetes use nearly doubled.

    A container packages together an application with all of its dependencies. That makes it much easier to migrate an application from, for example, a dev or test environment to a production environment, or from a private cloud to a public cloud, or from one public cloud service to another. All the leading cloud vendors offer container services, and increasingly, those services incorporate the Kubernetes container orchestration software. In fact, Forrester's Dave Bartoletti has gone far as to say, "Kubernetes has won the war for container orchestration dominance and should be at the heart of your microservices plans."

  • 5. Don't forget about security.

    bank vault

    These days, most organizations are less concerned about cloud security than they once were, but it remains a significant concern. The RightScale survey found that 77 percent of enterprises considered cloud security to be a challenge, and 29 percent said it was a significant challenge. It also noted that security is the number one concern for organizations that are new to the cloud, while experienced cloud users are more worried about cost.

    Before deploying cloud applications in production, organizations should understand the shared responsibility model and ensure that they follow best practices for data, application and network protection.

  • 6. Match the cloud service to the workload.

    Matching puzzle pieces

    It's easy to assume that the leading public cloud vendors are all the same. After all, their pricing often looks a lot alike and their services even seem to have similar names.

    However, experts say that it's a mistake to treat the public cloud like a commodity. Different vendors excel at different kinds of workloads, and pricing can vary widely due to small details in their pricing structures. In addition, cloud vendors are constantly adding new services and lowering their prices, so the service that is a best fit for a project this month might not be the best fit for a similar project next month.

    For these reasons, it's always best to evaluate cloud services on a case-by-case basis. By matching the workload to the best-suited service, organizations can maximize performance while optimizing cloud spending.

  • 7. Move toward software-defined infrastructure.

    Finger pointing to one node in a network

    While they may be moving some of their applications to the public cloud, most enterprises are going to continue running at least some workloads in their own data centers for the foreseeable future. RightScale's 2018 survey found that organizations run an average of 40 percent of their workloads in the public cloud. That's a sizable number, but it means that the majority of workloads are still running on infrastructure that organizations manage themselves.

    Many organizations are looking for ways to simplify configuration and maintenance of that infrastructure and unify the management of their in-house and public cloud resources. With increasing frequency, they are turning to automation and software-defined infrastructure (such as software-defined networking and software-defined storage) to help them meet those goals.

  • 8. Optimize your cloud spending.

    Clouds

    As previously noted, managing cloud spending was the number one cloud challenge for the experienced cloud users surveyed by RightScale. Overall, 76 percent of respondents said managing cloud spend was a challenge, and 21 percent said it was a "significant challenge." The report also noted, "Despite an increased focus on cloud cost management, only a minority of companies have implemented automated policies to optimize cloud costs, such as shutting down unused workloads or selecting lower-cost clouds or regions."

    When deploying applications to the cloud, organizations should make sure that they also have appropriate measures in place to manage the usage of cloud resources related to that application. If not, they can easily find themselves spending more than they anticipated on public cloud services.

  • 1 of

Improving Your Cloud Deployment: 8 Key Cloud Rules

  • 1 of
  • Clouds

    Improving Your Cloud Deployment: 8 Key Cloud Rules

    These best practices for cloud deployment can help organizations simplify and streamline the process of running production workloads on public cloud services.
  • architectural drawings

    1. Re-architect when you migrate.

    It's certainly possible to take legacy apps and simply move them to a public cloud service (a process often called "lift-and-shift" by cloud computing experts), but this approach might not result in any net benefit for the company.

    Many cloud consultants say you should start the process by considering the reasons why you want to migrate your legacy application. Are you hoping to achieve greater scalability? Are you trying to control costs? Is the migration part of a larger digital transformation initiative or a push towards greater agility?

    If the answer to any of those questions is yes, organizations might be better served by re-architecting or re-working an application, either before you migrate or after you have "lifted and shifted" the application to the cloud. This re-architecting definitely increases the work involved in migration, but if enterprise plans to continue using the legacy application for a long time, the time spent in re-architecting will pay off in terms of greater scalability, flexibility and all the other benefits offered by the public cloud.

  • metal gears

    2. Consider bare metal for legacy apps.

    Speaking of legacy app migration, you might want to consider a bare metal cloud service when moving existing applications to a public cloud. For those who don't know, a bare-metal server is a single-tenant server where virtual machines (VMs) are installed directly on the hardware (the bare metal) instead of in the operating system. (In others words, it's basically the kind of setup that IT departments used before virtualization and cloud computing became popular.)

    Bare-metal cloud services have some tradeoffs. On the one hand, they can offer faster performance than other cloud services for some workloads, and at least one vendor — IBM Cloud — will allow you to configure your bare-metal server to the exact specifications you need. That means you can exactly replicate your current data center in the cloud, which is helpful for lift-and-shift deployments. On the other hand, bare-metal isn't as scalable, flexible or inexpensive as traditional infrastructure as a service (IaaS). Still, it's worth investigating this option when migrating legacy applications.

    IBM and Oracle offer the best-known bare-metal cloud services. In addition, Amazon Web Services and Microsoft Azure have both announced previews of bare-metal services, and some smaller cloud providers also offer bare-metal options.

  • Computer screen with code

    3. Consider serverless for new apps.

    If you are deploying a brand new, cloud-native app, experts say you should consider a different type of public cloud service that is more or less the opposite of bare-metal: serverless computing. The name is something of a misnomer because serverless cloud services still run on servers. However, the developers writing the application don't have to think about or configure the cloud instances they are using. They simply write the code, and the cloud service handles the rest, configuring and scaling cloud instances as necessary.

    The serverless option is becoming increasingly popular, especially with developers building applications with microservices architecture or developers who work in DevOps environments. The best-known serverless option is AWS Lambda; other options include Microsoft Azure Functions, IBM Cloud Functions and Google Cloud Functions. Also, Oracle has an open source serverless project called Fn.

  • shipping containers with cloudy sky

    4. Containers make cloud deployment easier.

    Use of containerization tools like Docker and Kubernetes has skyrocketed recently. In the RightScale report, use of Docker increased 40 percent between 2017 and 2018, while Kubernetes use nearly doubled.

    A container packages together an application with all of its dependencies. That makes it much easier to migrate an application from, for example, a dev or test environment to a production environment, or from a private cloud to a public cloud, or from one public cloud service to another. All the leading cloud vendors offer container services, and increasingly, those services incorporate the Kubernetes container orchestration software. In fact, Forrester's Dave Bartoletti has gone far as to say, "Kubernetes has won the war for container orchestration dominance and should be at the heart of your microservices plans."

  • bank vault

    5. Don't forget about security.

    These days, most organizations are less concerned about cloud security than they once were, but it remains a significant concern. The RightScale survey found that 77 percent of enterprises considered cloud security to be a challenge, and 29 percent said it was a significant challenge. It also noted that security is the number one concern for organizations that are new to the cloud, while experienced cloud users are more worried about cost.

    Before deploying cloud applications in production, organizations should understand the shared responsibility model and ensure that they follow best practices for data, application and network protection.

  • Matching puzzle pieces

    6. Match the cloud service to the workload.

    It's easy to assume that the leading public cloud vendors are all the same. After all, their pricing often looks a lot alike and their services even seem to have similar names.

    However, experts say that it's a mistake to treat the public cloud like a commodity. Different vendors excel at different kinds of workloads, and pricing can vary widely due to small details in their pricing structures. In addition, cloud vendors are constantly adding new services and lowering their prices, so the service that is a best fit for a project this month might not be the best fit for a similar project next month.

    For these reasons, it's always best to evaluate cloud services on a case-by-case basis. By matching the workload to the best-suited service, organizations can maximize performance while optimizing cloud spending.

  • Finger pointing to one node in a network

    7. Move toward software-defined infrastructure.

    While they may be moving some of their applications to the public cloud, most enterprises are going to continue running at least some workloads in their own data centers for the foreseeable future. RightScale's 2018 survey found that organizations run an average of 40 percent of their workloads in the public cloud. That's a sizable number, but it means that the majority of workloads are still running on infrastructure that organizations manage themselves.

    Many organizations are looking for ways to simplify configuration and maintenance of that infrastructure and unify the management of their in-house and public cloud resources. With increasing frequency, they are turning to automation and software-defined infrastructure (such as software-defined networking and software-defined storage) to help them meet those goals.

  • Clouds

    8. Optimize your cloud spending.

    As previously noted, managing cloud spending was the number one cloud challenge for the experienced cloud users surveyed by RightScale. Overall, 76 percent of respondents said managing cloud spend was a challenge, and 21 percent said it was a "significant challenge." The report also noted, "Despite an increased focus on cloud cost management, only a minority of companies have implemented automated policies to optimize cloud costs, such as shutting down unused workloads or selecting lower-cost clouds or regions."

    When deploying applications to the cloud, organizations should make sure that they also have appropriate measures in place to manage the usage of cloud resources related to that application. If not, they can easily find themselves spending more than they anticipated on public cloud services.

In 2018, enterprises are shifting their attention from private clouds and hybrid clouds to public and multi-cloud environments.

According to the RightScale 2018 State of the Cloud Report, 92 percent of organizations use public cloud services, and 81 percent have a multi-cloud strategy. In addition, 71 percent of those surveyed plan to increase their public cloud spending by more than 20 percent this year, including the 20 percent of organizations that plan to more than double their public cloud spending.

But deploying applications in the cloud isn't always as easy as organizations hope it will be. While it's almost too easy for developers to spin up a cloud instance that they can use for development and testing, moving applications into production poses more challenges. IT needs to make sure those apps are going to perform at scale the same way they did during production. Plus, they'll need to worry about performance monitoring, security, costs and more.

And migrating legacy apps to the cloud is even more complicated that bringing a new cloud-native app into production. Existing applications typically have a lot of dependencies, and IT will need to make sure all those dependencies are replicated in the new cloud environment.

Fortunately, experts offer a lot of tips and best practices that can ease cloud deployment — whether you are launching new applications or migrating old ones. The following slideshow highlights eight cloud computing best practices for deployment.

Images from Pixabay

Submit a Comment

Loading Comments...