Cloud Services – Evolution and Future Trends

‍

Cloud services are no longer a buzz, they are existing fact. Small and large organizations are leading the revolution in the IT industry for almost a decade, some migrating small environments and conducting experiments while others deploying their entire production environments using this model.

It is commonly used to consider cloud services as a continue of the data center environment and in fact this is where the concept evolved, and yet, the differences between the traditional computing services, server farms, storage and even virtualization are fundamentally different from a true cloud. Let’s evaluate the differences:

In the past we used to call “cloud”, for any compute services containing the following characteristics as defined by NIST:

• On-Demand Self-Service

• Broad Network Access

• Resource Pooling

• Rapid Elasticity

• Measured service

When looking deeper into the various cloud service models such as IaaS (Infrastructure as a Service), PaaS (Platform as a Service) and SaaS (Software as a Service), we find that things are not always black or white: In some cases we come across services that we know for fact they are cloud services, we can’t always say these services include all the above characteristics.

A good example: File sharing services such as Dropbox. As a private customer we are not exposed to data that will enable us to measure the service (in terms of performance or in terms of billing vs storage capacity cost).

In case we choose to deploy a “private cloud” inside our organizational data center, based on infrastructure such as VMWARE, OpenStack or alike, we expect all the above characteristics in our on premise as well.

‍

Let’s differentiate between cloud and hosting service

In the current IT industry there are many companies offering compute services, on the range between cloud services and hosting services.

Hosting companies (or managed services), usually offers the customer the following capabilities:

• Compute environments – Such as physical servers (in case of special hardware requirements), virtual servers, storage and network equipment (Routers, Firewalls, VPN Gateway, etc.)

• Managed services – Such as web hosting for marketing or commercial web sites, email services, file sharing services and organizational systems such as CRM as a service.

• Backup and DR as a service.

• Managed support/IT services.

Hosting companies might offer the customer a capability to increase the number of servers and in some cases even to choose servers in data center abroad (in case we would like to allow access to the data/servers close to the end customer).

For SMB or enterprise organizations making their first move to the cloud, or for customers who wishes to outsource their IT services to external providers, there isn’t much of a difference between choosing hosting/managed service and choose public cloud service.

The differences between hosting and cloud services begins when trying to deploy entire environments in the cloud, based on architectures emphasizing service and platform (SaaS and PaaS), and less on infrastructure as a service (IaaS).

In this configuration, the system is developed based on dynamic scaling capabilities, environments deployed for a short amount of time, servers and infrastructure raised for specific purpose and stop to exist a few minutes after the entire process completes.

This model is called “Cloud Native Applications”, which allows us to avoid committing to pre-defined amount of infrastructure, specific task management, compatibility, server health check, etc., what is the role of each server or platform, in case they will be destroyed within a second? The infrastructure in this model is not important, only the service the system meant to provide.

Unlike hard-coded infrastructure management, there is a new concept – “Infrastructure as a code”. Environments are written as “recipes”, sent to the cloud provider using API’s, and environments are being created instantly and on the fly.

A few examples for the efficiencies of this model – A large American service provider destroys his entire IT environment in the cloud and deploys an entire up-to-date mirror environment within a few minutes instead of updating each and every server. A different American service provider increases the amount of servers automatically in advanced before peak hours, as a result of applicative monitoring, and after peak hours, all the new servers vanishes.

This change is not a magic, but a result of cloud efficient planning of systems and applications, training dedicated teams with the relevant capabilities, understanding the cloud provider’s services, billing, prioritization and the constant changes in the management interfaces.

Process of migrating systems to the public cloud

Infrastructure as a Service (IaaS) allows organizations to perform “Lift & Shift” (copying existing systems to the cloud with minor changes) from the on premise environment to the public cloud, as part of migration processes to the cloud.

Most organizations will quickly find out that the “Lift & Shift” strategy is indeed easier as first stage, but in the long term it is a very bad economical decision, with the same challenges that organizations struggle with today: waste of hardware resources, un-optimized operating system and running code on servers, distributed management difficulties, etc.

At later stages, organizations who migrated systems to the public cloud, begin to perform tuning to their cloud environments by measuring resource usage, for their virtual servers, adapting the proper VM instance type for the actual use in terms of CPU/memory/storage.

Below is an example from AWS presentation about the evolution organizations pass migrating to public cloud in terms of cost:

‍

‍

The future is here

Today, startup companies and enterprise organizations are developing applications for the cloud, agnostically to the hardware or infrastructure underneath.

In past, organizations used to migrate from developing on physical servers to virtual servers, and today, organizations are migrating from developing monolith applications to micro-service based applications and even Serverless applications (code running inside a managed compute environment, without the need to manage infrastructure resources such as servers, storage, etc.)

Changes in the development world is crucial to understanding what is cloud service, due to the fact that today, we are less relying on NIST definition of cloud service, and providers offering Infrastructure as a Service (as most hosting providers) and today cloud service is characterized by the following capabilities:

• Collection of API’s

• Billing capability of services/resources by their actual usage

• Services managed using API (such as the ability to provision, decommission, start/stop, etc.)

The bottom line

Today there are many providers who wrap VMWARE infrastructure with friendly user interface, allowing the customer to choose the VM type (in terms of CPU/Memory) and the number of servers the customer would like to consume, but it is not scale enough and it doesn’t allow the customer the flexibility to scale-up or scale-down to hundreds of servers automatically, and within seconds over multiple geographical regions.

Cloud provider who supports “Cloud Native Applications” enables the customer to connect his automation mechanisms, build and deployment processes using API’s to the cloud provider’s infrastructure, in-order to allow provisioning/changing compute environments, deploy micro-services based systems and even allowing the customer to deploy and run Serverless infrastructure.

The next time you are considering a service provider, whether using a hosting provider (sophisticated as he might be) or using public cloud provider, with capabilities of running hundreds of servers over multiple geographic regions within minutes, hands-free, supporting micro-services and Serverless applications, with API connectivity capabilities, simply present the service provider your requirements, and choose the most suitable service provider.

‍