Postgres API for SQL and “Cassandra” for no-SQL databases Free hybrid IT architectures from dependencies
22.02.2022A guest post by David Walker *
The IT architecture of data centers has to evolve in view of the edge dynamics triggered by IoT and 5G. It wonders how. Because this evolution will bring with it a wealth of innovations.
With Hybrid Cloud, users strip manufacturer and provider dependence.
In today’s IT, there are numerous different application architectures: central data centers; central architectures supplemented with regional or local IT resources or completely distributed resources. The question is how the architecture of data centers must evolve in the face of the edge dynamics triggered by IoT and 5G to ensure that the data is available to end users wherever, whenever and however they need it.
We are currently experiencing a new development spurt in the data centers. The trigger is the transition from the first to the second generation of cloud implementations; the fuel is provided by the increasing spread of 5G and IoT applications.
The first generation was about internalizing the use of the cloud and “somehow” identifying databases and applications that were suitable for this new environment. Now that the use of the cloud has become established, user companies are gradually finding out its downsides: they are tied to their cloud provider. And because the IoT increases the volume of data, the costs also increase.
In an attempt to overcome the monopolies of cloud providers, some companies place certain application classes in certain clouds, for example “Azure” for “Office”, AWS for OLTP and “GCP” for analytics. This reduces some of the dependence on cloud providers. But users are still dependent on a cloud provider for each application class.
The development of a second generation of cloud usages will be characterized by three main features:
- 1. The use of applications, in particular databases, distributed among several cloud providers.
It will not be enough to simply distribute applications to several clouds – they must rather be distributed both via clouds and via their own in-house IT capacities.
In this way, companies can bring processing and storage closer to the user and move it to the most cost-effective platform for the specific type of workload. Companies that use provider-specific databases from cloud operators will have to rethink this approach. Users who want to migrate their stateless, cloud-native microservices across cloud providers need a data layer and, in particular, a database that is supported across the cloud.
- 2. The use of three data centers in different regions – in three instead of two.
While it was previously customary to operate two data centers – one active and one passive – in the future, the use of distributed databases at three locations should replace this concept. If one of them fails, the services are still retained. Operating in Frankfurt, London and Dublin is no more difficult than operating at two of these locations. This is not much more complex, but at least much more robust. And it can be easily implemented using the technologies that are now available.
The three-way approach also better supports compliance with legal regulations and data security. This makes it possible, for example, to store data exclusively in one of the three centers in the USA, the EU and the rest of the world in order to comply with legal requirements, and at the same time to store them in a single global database.
- 3. Transition to truly agnostic APIs for data storage.
I predict that the “Postgres API” for SQL databases and “Cassandra” for no-SQL databases will prevail, while the programming language “S3” is increasingly being used for block storage. In fact, this amounts to a unification of the APIs for data storage. This is important because it removes a barrier that makes scaling difficult and prevents companies from switching between different platforms at the data level.
David Walker: “The question is how the architecture of data centers must evolve in the face of the edge dynamics triggered by IoT and 5G to ensure that the data is available to end users wherever, whenever and however they need it.“
(Peter Duce Photography)
The commoditization of the data center is already underway. These three trends are the inevitable result of the user’s desire to scale cost-effectively without having to commit to a provider. The providers will not give up their lock-in advantages without a fight, but ultimately the approach that promises users more independence will prevail.
* David Walker is Field CTO at Yugabyte in Europe, Middle East and Africa.