As the growth and use of virtual SANs continues to grow in enterprise data centers, demand for projects and implementations on network-oriented (L3) has increased. One of the key components of a software-defined data center is the design and implementation of Virtual SAN that must meet the same requirements and requirements as other components of the software-defined data center.
Designing virtual SAN implementation in Layer 3 networks requires sufficient planning and understanding of virtual SAN Design capabilities and network infrastructure to be optimized and implemented in the most efficient manner. flexible and efficient to the full scale.
Virtual SAN extends to the concept of software-defined data centers by grouping distributed resources (storage devices) and managing them as a single entity (data archives), achieved by relying on sufficient network infrastructure to provide a seamless connection between virtual SAN cluster nodes.
While developing enterprise-level hyperconverged infrastructure projects for virtual SANs that support the deployment of similar components (hosts), it has a network infrastructure capable of meeting connectivity requirements and providing connectivity for up to 100,000 hosts. Currently, the network infrastructure must be flexible and capable of supporting the transfer of virtual SAN cluster components as needed.
Imagine a network cable breakdown to connect the nodes
See the above statement on how to install a network cable on each shelf where you can download and connect a virtual SAN host. It must be a highly predictable, widely scalable, and durable type of network to support storage and other traffic in SDDC.
This year for VMworld, I partnered with Brocade to develop a validated and tested design and deployment guide for Virtual SAN through a large-scale network infrastructure based on Brocade network devices.
Fabric IP Brocade architecture, which can grow significantly in and among several data centers, has been well received in the industry due to its automated and transparent operation. Brocade offers retrospective leaf architecture in which servers connect to interconnected leaves through a plug that provides a high-performance, deterministic and high-performance network.
Traffic between leaves and thorns is based on the L3 IP sublayer - using multiple lanes at the same cost to use all available connections and BGP to provide routing. Relatively simple, right? - Yes! If you're a network expert!
Virtual SAN implementation.
We decided to endorse and recommend the proposed project for this type of architecture configuration, and we created a network to show virtual SAN distribution on network infrastructure that is actually channelled to 5 data center shelves, each rack having a pair of reed (ToR) switches: two pole switch
Virtual SAN design is designed to work natively on L3-directed networks and is configured with high availability on the shelf with the use of virtual SAN error domains to prevent service interruption or data access by rack failure. With five shelves (5 virtual SAN error domains), the number of simultaneous shelf failures that can be tolerated by the virtual SAN infrastructure is two (2N + 1 formula).
In addition, all shelves are interconnected via the IP Fabric in the same way and do not need to be on adjacent tiles, even in the same DC room! And based on scalability, even just by adding more shelves to the diagram above, it covers over 850 servers. This is what I call flexibility, scalability and consistency!
Learn more about designing virtual SANs and consulting experts
Based on the validation of our solution, we wrote a white paper on the elements to consider during design and how we deploy. Be sure to stop by the Brocade booth at VMworld and get a live demo of everything and take advantage of our experience by designing and implementing a virtual SAN on a Level 3 network.