Best Practices for Hybrid Cloud Connectivity
by Toshal Dudhwala
Cloud architectures allow automated, on-demand delivery of applications and services, flexibly deployed across large, cost-effective resource pools, whether on-premises or from service providers. The automation and delivery of on- demand services are driving previously unimagined business agility and a new generation of business applications and revenue opportunities. Organizations must often weigh the trade-offs of additional complexity and overhead of an on-premises private cloud, with the limited controls and compliance risks inherent with public cloud providers.
Hybrid cloud is rapidly becoming the optimal solution for most enterprises. It delivers cost-effective shared services between a private cloud and public cloud provider—Virtual Private Cloud (VPC) such as Amazon Web Services (AWS), Microsoft Azure Cloud, or Google Cloud Platform (GCP). In its simplest form, a hybrid cloud seamlessly connects on-premises private cloud infrastructure to a public cloud provider, allowing the exchange of data and applications between locations. Bridging an application network, such as a three-tier web application, across multiple sites and providers, with the appropriate network and security policies, and the same ease of orchestration as if the entire app was running on-site, can be a complex challenge to overcome. There are many reasons and use cases for why an enterprise would adopt a hybrid cloud model:
- Cloud bursting or application elasticity – Many enterprises provision their on-premises application workloads for baseline or median capacity to reduce public cloud outsourcing costs, and to keep local resources fully utilized. During peak times (e.g. holiday
season, special events, promotional windows), this median capacity won’t deliver optimal performance. In
order to address this peak load, many enterprises burst excess capacity to the public cloud to limit additional IT spending to precisely when it is needed.
- Architectural flexibility – Enterprises frequently split their applications into two or more components and host some on-premises for data privacy and others in the public cloud VPC. Hybrid cloud can also be used for new application development, where only temporary compute resources are needed while new applications are being developed.
- Workload localization – Hybrid cloud may be used when enterprises are required to bring the application
close to end users due to local laws
or simply to improve the end-user experience. In this case workloads may be distributed across multiple cloud providers in multiple geographies.
- High availability and resilience – Enterprises need to design for the worst-case scenario. A public cloud option provides them backup for their on-premises datacenter. Rather than waiting for disaster to strike, a certain amount of live workloads and capacity, along with mission-critical data is usually running in parallel to assume capacity without interruption.
There are a few common ways in which enterprises connect their on- premises datacenter and a public cloud provider. The most common way is a VPN connection between the private datacenter and the cloud provider. The public cloud provider has a VPN gateway where the enterprise terminates the IPSec tunnel from their datacenter or branch of ces (WAN). For software- defined networking (SDN) automation and cloud orchestration, the hybrid cloud link is usually a VXLAN overlay virtual network, making it transparent to the application workloads whether they are running remotely or on adjacent CPUs.
Challenges with hybrid cloud connectivity and automation
Providing such a seamless level of integration across virtual networks and sites, with transparency that avoids application redesign and allows for full flexibility, does present some challenges:
- Scalability – A typical enterprise may have dozens or hundreds of virtual application networks. As the number of VPCs grows, it creates challenges for enterprises to maintain and manage many secure IPSec tunnels to the various VPCs.
- Security and segmentation – An enterprise typically has many internal business groups with various policies. This requires a strict segmentation
of the workloads, or logical isolation across various hybrid cloud networks. Public clouds require additional capability to enforce these policies compared to private cloud architectures.
- High availability – Continuous availability of workloads under all conditions and scenarios is frequently a mission-critical business requirement. The connectivity to the public cloud workloads also needs to be considered due to possible failures.
- Application visibility – Visibility to application performance and potential security anomalies is required for optimal service delivery. A frequent challenge is how to get the insight into the application workloads coming in and out of a public cloud as easily as ones on-premises.
- Centralized policy and management – A typical large enterprise uses multiple cloud providers and the connectivity and management of applications, services and workloads are likely different
for each provider, adding additional complexity and orchestration concerns.
Things to look for in Hybrid Cloud Connectivity solution
- Unified policies – The Enterprise should be able to extend the same user and application policies to the public cloud instances as it applies to on-premises infrastructure and remote of offices.
- Scalable networking – The traditional “hairpin” model where all VPC traffic must traverse via on-premise datacenter gateway is not scalable. A VPC should be able to directly connect to any Enterprise sites including branch as well as data center.
- Resiliency – For the enterprise sites with multiple transports available (both internet and MPLS VPN), the solution should be able utilize the best available path considering both cost and performance requirements to optimize network access and WAN costs.
- Context-aware security – The solution should apply same degree of security policy automation, along with virtualized security appliances, as has traditionally been found only in on-premises deployments.
- Application Aware Networking – Use of the best cost-effective path to VPC application which is compliant to business policies and service level agreement
- Operational efficiency –The same degree of SDN automation that was previously only available for private clouds should easily be extended to multiple hybrid cloud deployments.
To learn more, stop by Nuage booth at ONUG Spring 2017 in San Francisco.
Toshal Dudhwala is responsible for leading Product Solution and Partnership for Nuage Networks SDN wholly owned subsidiary business unit of Nokia (former Alcatel-Lucent). In this role, he actively engages with Enterprises and Service Providers customer as well as Technology Alliance Partners to solve problems through new and innovative services based on Nuage Networks SDN and SDWAN solution. Toshal has experience working directly with partners and end-customer on every aspect from initial executive level business case development, to solution definition and solution implementation.
Prior to joining Nuage, Toshal was leading Product Management effort for NEC’s ProgrammableFlow SDN solution as well as Ixia’s Software Defined Networking (SDN) and Data Center solutions. Toshal has over 14 years of experience in the data communication industry, specializing in emerging technology such as IP/MPLS and SDN. He holds a degree in Master in Telecommunication Engineering from Monash University Melbourne, Australia.