Are you a community engineer who’s tinkering with community programmability, however want extra assist implementing your imaginative and prescient for mass provisioning and/or observability?
Or, perhaps you’re a software program developer who’s thinking about utilizing your programming, API, database, and visualization abilities in a brand new means with networking, however want product perception to know what’s impactful?
If both of those are true, you is perhaps thinking about some background on how we used DevOps fundamentals within the CiscoLive Community Operations Heart (NOC).
A pair months have handed since CiscoLive 2022 in Las Vegas, so it appears a very good time to replicate. I had the privilege to take part once more within the Cisco Stay NOC – specializing in automation, observability, and community programmability. I’ve served multi-purpose roles of speaker and event-staff/NOC for a few years. It’s one in all my favourite actions in the course of the yr.
Operating a safe, continuous, high-performance community
Preparation for the June Las Vegas occasion began quickly after the New 12 months. We start with our base necessities of operating a safe, continuous, high-performance community that showcases product capabilities and advantages. We collect product steerage and new-release intent from the assorted product groups and map that towards well-developed community infrastructure designs we’ve used year-over-year that embrace modularity, layered safety, service resiliency and virtualization.
CiscoLive Community Topology
NOC subject material consultants
The final in-person occasion for the US was CiscoLive 2019 in San Diego. Since then we’ve realized many product adjustments. Individually, I used to be to train new streaming telemetry capabilities that had been evolving a number of years earlier.
The NOC has a number of subject material consultants throughout the corporate in Buyer Expertise (CX) [includes TAC], Engineering, IT, and Gross sales. I’m coming from the Engineering facet in our Developer Relations crew (DevNet). Whereas accustomed to most of our networking applied sciences, I specialize within the programmability, APIs, databases, dashboarding, virtualization and cloud-native applied sciences wanted to attach our industrial merchandise along with open-source options.
To get the community monitoring facets solidified we depend on a standard DevOps mannequin the place a developer – me – pairs up with subject material consultants in every of the assorted IT domains. I met with area consultants like Jason F on routing, Mike on safety, Chris on wi-fi, Richard on switching and others to make sure correct operational protection.
Acquainted instruments FOR IT Service Administration
We clearly use our industrial merchandise like Cisco DNA Heart, Cisco Prime Community Registrar, Cisco Id Companies Engine (ISE), Intersight, the Wi-fi LAN Controller, Meraki Dashboard, Crosswork, ThousandEyes, and the assorted SecureX options. We complement these with different acquainted instruments of an IT Service Administration setting, like VMware vCenter, NetApp ONTAP, Veeam, and specialised instruments like WaitTime. We continued our occasion design by figuring out the configuration administration and provisioning features we wish and take into account the efficiency, fault, and telemetry metrics we hope to realize. The performance every software gives is mapped towards these intents and gaps are recognized. At this level we take into account workaround choices. Is supplemental info obtainable instantly by APIs? Can we develop a proximate analog by combining information from a number of sources? Do we’ve to drop again to legacy strategies with SNMP and/or CLI scraping?
When we’ve a stable understanding of the necessities, merchandise for use, the administration software capabilities and the gaps, we frequently use open-source options to fill within the wants. For the reason that area consultants are normally targeted on their domain-specific administration instruments, the open-source merchandise, programming, and API orchestration actions amongst all of the instruments are the place I focus my energies.
Extra streaming telemetry
As I discussed, I needed to make use of extra streaming telemetry at this yr’s Cisco Stay occasion. We did. Not solely did we use gRPC dial-out with some dial-in, but in addition NETCONF RPC polling for effectivity in some instances. SNMP was not utilized by ANY of our tailor-made collectors – solely with a number of the industrial purposes. Because the programmability SME, I meant to make use of YANG fashions with gRPC streaming telemetry or NETCONF RPC, the place obtainable. I frolicked reviewing Cisco’s YANG Mannequin Repository and used YANG Suite to check the fashions towards a staging setting.
For instance, the ‘Web Site visitors Quantity Dashboard’ confirmed what number of terabytes of visitors we exchanged with the Web.
CiscoLive Web Site visitors Quantity Dashboard
We liken it to the ‘Tote Board’ that celeb Jerry Lewis used to indicate what number of thousands and thousands of {dollars} had been raised for the Muscular Dystrophy Affiliation charity in the course of the Labor Day telethons from 1966 to 2014.
Creating our dashboard
Our dashboard was created by extracting stats from the sting ASR1009-X routers offering our 100 Gbps hyperlinks to the Web. A customized NETCONF RPC payload, like the next, was requested each 10 seconds.
<filter xmlns="urn:ietf:params:xml:ns:netconf:base:1.0"> <interfaces xmlns="http://cisco.com/ns/yang/Cisco-IOS-XE-interfaces-oper"> <interface> <title/> <interface-type/> <admin-status/> <oper-status/> <last-change/> <pace/> <v4-protocol-stats> <in-pkts/> <in-octets/> <out-pkts/> <out-octets/> </v4-protocol-stats> <v6-protocol-stats> <in-pkts/> <in-octets/> <out-pkts/> <out-octets/> </v6-protocol-stats> </interface </interfaces> </filter>
Instance – NETCONF RPC payload to acquire environment friendly interface stats
As soon as acquired, the data was parsed into an InfluxDB time-series database, then Grafana was used to question InfluxDB and produce the dashboard visualization as above. Utilizing this methodology allowed us to be extra granular concerning the particular stats we needed and diminished the necessity for filtering on the collector. We look ahead to enhanced capabilities in defining particularly filtered sensor-paths for gRPC dial-in/-out for future deployments.
Persevering with the DevOps theme of collaboration, our WAN SME was eager to get optical transceiver energy ranges on the 100 Gig and 10 Gig Web hyperlinks. We mentioned how If the transmit or obtain energy ranges dropped beneath optimum thresholds, then efficiency would endure, or worse – the hyperlink would have a fault.
Sadly, the instrumentation was not supplied on the particular person lane [optical lambda] degree we needed, so we had a niche. Researching work-arounds resulted in a fallback to CLI-scraping of ‘present transceiver’ instructions. Whereas this was suboptimal, I used to be capable of provide you with a programmatic methodology to ballot the unstructured information and inject it into InfluxDB utilizing Python. The Python schedule library was used to automate this course of each 2 minutes. Grafana dashboard templates had been created to render information queries to our most well-liked visualization. The outcomes had been useful as we may see every lane of all optical transceivers and their transmit and obtain energy ranges.
Grafana dashboard of Optical Transceiver Energy Ranges
Aligning with DevOps
One other DevOps-aligned scenario was managing observability with a number of wi-fi environments. The principle convention community, adjoining lodges and keynote space had been managed by totally different clusters of Cisco Wi-fi LAN Controllers (WLCs). The 4 Seasons resort had Meraki Wi-Fi 6-enabled entry factors managed by way of Meraki cloud. After consulting with the wi-fi SMEs to get their deep insights, expertise, and intent, I reviewed the present state of the administration instruments and portals. To fill in gaps and lengthen present capabilities, I researched the instrumentation and telemetry obtainable within the WLC and Meraki Dashboard. There was a combination of NETCONF and REST API strategies to contemplate and we relied on Python scripting to assemble information from totally different sources, normalize them and inject into InfluxDB for Grafana to visualise. The outcomes clarified utilization and useful resource distribution.
Converged WLC-Meraki Wi-fi Shopper Dashboard
Converged WLC SSO Pair Well being Standing Dashboard
Professional Mode WLC Wi-fi Community Controller Daemon (WNCd) Well being Monitor
Placing all of the instruments and assortment scripts collectively gives us a number of observability dashboards; some we present on the video show wall the place you may see what we do!
Many different examples of DevOps collaboration and tailor-made observability might be shared, however this weblog would scroll a lot, for much longer! In the event you’re thinking about a number of the Python scripts and Grafana dashboards developed for the occasion, try this Github repo – you may be capable to repurpose a number of the effort on your particular wants!
Hopefully this weblog impressed you to collaborate in a DevOps style; in case you’re a community engineer – hunt down a developer to help make your imaginative and prescient a actuality; in case you’re a developer – hunt down a community engineer to get extra insights on how the merchandise work and what instrumentation and metrics are most significant.
For extra info of NETCONF, gRPC, telemetry, examine these studying sources out:
Observe me on Twitter for the subsequent occasion – I’d be completely happy to speak to you!
We’d love to listen to what you suppose. Ask a query or go away a remark beneath.
And keep related with Cisco DevNet on social!
LinkedIn | Twitter @CiscoDevNet | Fb | YouTube Channel
Share: