Hybrid cloud adoption is becoming the de facto standard in most enterprises’ technology architecture. While the public cloud is a core component of a hybrid cloud architecture, the private cloud platform is gaining importance as well. Concerns about data security and compliance with data residency regulations have forced enterprises to shift to the private cloud platform. Cloud hyperscalers need to move quickly in addressing this shift by strengthening their private cloud offerings or risk losing customers in the long run.
Registered users can unlock up to five pieces of premium content each month.
Log in or register to unlock this Insight.
Need for a More Deliberated and Calculated Approach to Hybrid Cloud Infrastructure
|
NEWS
|
A common challenge in hybrid cloud architecture is the disparate nature between legacy infrastructure and new business applications. The proliferation of Artificial Intelligence (AI), for example, has increased the need for a robust private cloud platform. A more deliberate and calculated approach to implementing a hybrid cloud platform is needed. IBM calls this approach the “hybrid-by-design” framework. This framework aims to facilitate the translation of business priorities into key architectural decisions.
Central to the hybrid cloud architecture is the private cloud platform component. This is a growing space that both technology vendors and cloud hyperscalers are moving into. Among the more prominent private cloud providers is VMware with its VMware Cloud Foundation platform that allows platform engineers to create private cloud environments within existing data centers, Amazon Web Services’ (AWS) Virtual Private Cloud, which includes virtual provisioning of EC2 and database storage instances, and Microsoft’s Azure Dedicated Host for Azure Virtual Machines (VMs).
Cloud Hyperscalers Risk Losing Enterprise Customers Looking to Deploy Private AI
|
IMPACT
|
Focus on the private cloud has grown in prominence, along with the demand for Generative Artificial Intelligence (Gen AI) solutions. While a large majority of enterprises will be using public cloud platforms to develop Gen AI solutions, enterprises in industries that are highly regulated such as financial services, government agencies, and healthcare will have to look at private cloud platforms or private AI to be compliant with industry-specific regulations (e.g., the Health Insurance Portability and Accountability Act (HIPAA) and the General Data Protection Regulation (GDPR)). In this respect, cloud hyperscalers are slower to offer enterprise Gen AI solutions in a private cloud environment compared to technology providers such as VMware. The introduction of VMware’s Private AI Foundation with NVIDIA gives large enterprises the choice of deploying Gen AI solutions in a secure environment without exposing confidential data over a public cloud platform.
Cloud hyperscalers are also facing challenges over data integration, requiring enterprises to be fully embedded in their public cloud platform to benefit from true cloud cost savings. This has allowed data integration vendors such as Informatica, Domo, Snowflake, and Databricks to provide an alternative in connecting data silos spread across the enterprise. Over time, this will relegate cloud hyperscalers’ role to a compute and data storage provider, while enterprise software and data management technology providers take over the responsibility of solving data management challenges, resulting in potential revenue loss for cloud hyperscalers.
Introduction of Small Language Models Could Be the Catalyst for Private Cloud Dominance
|
RECOMMENDATIONS
|
Cloud hyperscalers are leading the way in terms of hosting Gen AI solutions. However, enterprises looking to leverage Large Language Models (LLMs) like Llama 3 do not necessarily even need to use Amazon Bedrock or Azure AI on a public cloud platform. To compete with legacy private cloud providers, cloud hyperscalers need to leverage their dominance in Gen AI by introducing innovative Gen AI capabilities that can be hosted on their private cloud infrastructure. An example of this would be offering Small Language Models (SLMs) hosted in a private cloud. Microsoft has introduced the Phi-3-mini, an SML that is trained on smaller amounts of data compared to LLMs. While not currently offered in a private cloud environment, the introduction of SMLs could be a significant step in establishing a leadership position in the private cloud market. Introducing SMLs into a private cloud infrastructure means that enterprises would not need to worry about data privacy concerns. SMLs are more cost-effective and optimized to work on limited resources, perfect for a private cloud infrastructure.
From an enterprise user perspective, when choosing a private cloud provider, it is important to have a clear idea of which private cloud architecture is the right choice for the business, as this will determine the Quality of Service (QoS) and investment cost for deployment. The three main private cloud architectures are:
- On-Premises Private Cloud: Traditional on-premises infrastructure that will be managed by the internal Information Technology (IT) team. This offers the highest level of control and security, but is also costly to maintain and operate due to upfront and recurring expenditures.
- Virtual/Hosted Private Cloud: Offered by public cloud providers with dedicated resources specific to the enterprise. All resources are separated from a virtual layer perspective, but still share the same physical hardware with other enterprises.
- Managed Private Cloud: A single-tenant environment managed by a cloud service provider. The physical hardware infrastructure can be hosted in a third-party data center or within an enterprise’s own data center. Examples of this architecture are AWS Outpost and Microsoft Azure Stack.
