Skip to content 
Datacenter Automation and Network Evolution deliver extraordinary outcomes when ‘what a network works for’ aligns perfectly with ‘what a network is built of’
Networks have undergone a paradigm shift. Almost every modern business relies on networks for tasks that were implausible just a few years back.
Networks have become intelligent, software-driven, and fluid enough to take care of agile business needs. In the current application-economy and digitally-awake world, networks construe a force that cannot afford to snooze or let an opportunity slip by. These areas become all the more compelling, given the changes that data center automation is ushering in for IT environments.
Uptime, downtime, latency– these are among the many metrics that are no more out of the network’s ambit anymore. These parameters are inch-tapes that measure a network’s agility and effectiveness every nanosecond.
Additionally, networks also have to find a new rhythm with today’s data centers that are evolving rapidly. Case in point – A bulk of organizations have moved to the leaf-spine architecture. Flocking applications, on-demand workloads, mobility of virtual devices, cloud integration, scale, & customer SLAs have posed some serious challenges for an enterprise’s network.
One of them is disguised under one disruptive technology advancement as well. Look at the introduction of overlay technologies, such as VXLAN, that have helped IT teams to handle a lot of complexities. They work splendidly. But the big question is, how well is an underlay overlay integration done in your data center?
Administrators have to be able to deploy and expand workloads anywhere in the datacenter. Many traditional limitations of layer 2 networks are obscured by overlay technologies allowing them to expand without affecting the availability of existing services – and still providing secure segmentation and performance isolation. However, overlay technology poses several challenges:
It’s hazy – Underlays and Overlays work independently. This lack of mutual awareness keeps NetOps teams searching for network blind-spots between the layers and, often, unable to correlate & find out the reason for performance or network outages. The integration of underlay and overlay networks demands a feedback loop in the network to provide the much-needed correlation.
Inventory management isn’t easy – While overlays make life easier, their presence does not keep NetOps teams absolved of the need to maintain and configure the IP and VLAN constructs. Maintaining manual records of overlapping IP address space, network segregation, VLAN to VXLAN mapping, and other resources involved burdens some already-existing documentation. While SDN opened up new avenues, it still co-exists with legacy infrastructure in many organizations – this translates into even more information that has now swelled up for Network teams to handle. Datacenter management involves multiple teams including network and compute teams. With overlay being a huge part of virtual workloads, there exists an ambiguity on the ownership of this layer causing further confusion. There is a need to streamline these processes in managing the underlay and overlay layers in a Datacenter.
Manual provisioning – Whenever a new workload is planned in a data center, it is common for NetOps teams to start creating a configuration template – manually. While home-grown scripts can be helpful, they fail to equip everyone with a larger perspective of the network and its topology. Workload mobility, Workload migration, virtual-to-virtual, virtual-to-physical or physical-to-physical connectivity, network overlay end-point traffic policies, and much more – all this tosses enough on the plate to encumber and complicate custom scripts and open-ended automation tools.
Scale – Unlike other domains, Datacenters are dynamic. Workloads are spun on-demand, pushing network expansion at a rapid rate. Monitoring is, hence, an important aspect of underlay-overlay integration. The tons of metrics collected to analyze the health of the network require the monitoring platform to auto-scale on-demand to meet the growing requirements of the network. Monolithic tools and custom scripts fail to scale at this pace and level. They cannot meet the growing demands of Datacenter networks with their limitations and small radius control.
Introduce underlay overlay integration with Anuta ATOM
Anuta ATOM provides some much-needed relief with and around underlay-overlay integration within data centers. It aligns very well with the new context and needs of data center automation.
Correlate underlay & overlay metrics – ATOM acts as a robust collector of multiple datasets such as SNMP, SNMP traps, Streaming telemetry, & Syslog. With a microsecond-level of granularity, ATOM is capable of providing deep insights both from the underlay and the overlay networks. With its sharp Closed-loop automation capabilities, any changes in the underlying network such as congestion in the links, packet loss, Jitter, or even an outage in the physical network can get easily correlated to performance degradation in the overlay networks.
Discover your data center – Be it SDN or legacy infrastructure, ATOM can assimilate it all. ATOM can integrate into SDN Controllers such as Cisco ACI or Juniper Contrail or onboard other multi-vendor devices with impressive ease. While ATOM supports internal IPAM and VLAN management, it also integrates into IPAM devices such as Infoblox &Bluecat, thus, relieving NetOps teams of time-draining chores of record-keeping. ATOM helps to empower the NetOps teams with all the resources they need by their side to handle their dynamic data center environments.
Data center Automation – In certain scenarios, where a virtual network is planned to onboard a new application, ATOM arms you with the necessary abstraction to provision a multi-vendor infrastructure with the necessary VLANs connecting endpoints, mapping it to the VXLAN tunnels between the endpoints and Layer-3 VRF. The IP address for the layer 3 can be taken from external IPAM systems or internal IPAM module of ATOM. One can also specify the distributed anycast mac address for leaf switches, the BGP AS number, and the spine-switch to exit the Pods. ATOM ensures resource-availability followed by apt generation of vendor-specific commands to create VLANs, VNI, L3 information, VLAN-to-VXLAN mapping on the leaf switches, ARP suppression and other relevant configurations. Execute datacenter automation using ATOM’s service modeling & low-code workflow automation.
Scale as you need – With a modern software stack and strong microservices architecture, ATOM easily and effortlessly scales to meet the growing requirements of any data center environment. Each component – including the time-series database within ATOM- is developed as a microservice, allowing it to scale on-demand to ingest tons of metrics. ATOM’s scale also aids the feedback loop to correlate the underlay and overlay metrics it collects as a monitoring solution in the data center environment.
While overlays have transformed how networking is perceived, they can only be as reliable and effective as the underlays that serve them. That unseen wall between these layers acts as a huge, but neglected, barrier, and there are not many solutions that are capable of addressing this issue.
Embrace Anuta ATOM – a comprehensive multi-faceted automation solution that reads into these blind spots between the underlay and overlay layers and provides you the single-source-of-truth for your network. Embrace data center automation through underlay overlay integration with a new level of confidence.
