Last year Arkin burst onto the scene offering a solution that focused on virtual and physical deep network analytics. Arkin was recognised at VMworld 2015 by nearly taking out the best of show and fast forward twelve months, Arkin was acquired by VMware with the product later rebadged as vRealize Network Insight. One of the products main strengths that attracted VMware into making the acquisition was it’s tight integration into NSX by way of a simple and intuitive user interface that lets admins easily manage and troubleshoot NSX while offering best practice checks that can guide users through VXLAN and firewall implementations and alert them to any issues in their design and implementation of NSX.
Arkin removes barriers to SDDC adoption and operation by providing converged visibility, and contextual analytics across virtual and physical, an ability to implement newer security models such as micro-segmentation, and by ensuring application uptime, while letting IT collaborate better. The platform helps IT organizations plan, operate, visualize, analyze, and troubleshoot their complex software-defined data center environments.
As vRealize Network Insight the key benefits are:
- East-west traffic analytics for security and micro-segmentation design
- Control and tracking to meet audit and compliance requirements for virtual distributed firewalls
- 360 Overlay-underlay visibility and topology mapping
- Extensive 3rd party physical switch integrations
- VXLAN to VLAN logical path mappings
- Advanced NSX Operations Management
- Natural language search and enhanced user experience for rapid troubleshooting
What I was surprised to find when I was able to dig into the product was that it offered more than just Network insights…in fact it offered surprisingly deep analytics and metrics for Hosts and Virtual Machines that rival most similar products out on the market today.
To install Network Insight you download two OVA’s from MyVMware and deploy the two appliances into vCenter. It’s got an interesting setup that’s shown below and after deployment you are left with two appliances, a Platform, and a Proxy that have the following specifications.
- 6 CPU cores (reservation 3072) Mhz
- 32 GigaBytes RAM (reservation)
- 600 Gigabytes HDD (thin provisioned)
- 2 CPU cores (reservation 1024 Mhz)
- 4 Gigabytes RAM (reservation 4GB)
- 100 Gigabytes HDD (thin provisioned)
A note before continuing…only Chrome is supported as a browser at this stage.
You start the install by deploying the Platform appliance…once the Platform OVA is deployed and the appliance VM settings have been configured you can hit the IP specified in the OVA deployment process and continue the installation.
After the license key has been validated you are then asked to Generate a shared secret that is used to pair the Platform with the Proxy appliance.
From here you can initiate the deployment of the Proxy appliance. During the OVA deployment you are asked to enter in the shared key before continuing to configure the appliance networking and naming. As shown below, the configuration wizard waits to detect the deployed Proxy appliance at which point the installation is complete and you can login.
The default username name is admin@local with a password of admin.
When you login for the first time you are presented with a Product Evaluation pop up letting you know you are in NSX Assessment Mode and that you can switch to Full Product Mode at the bottom right of the window. NSX Assessment Mode is an interesting feature that looks like it will be used to install Network Insight as part of an on boarding or discovery engagement and produce reports on what is happening inside an NSX environment.
In either mode you need to register at least one vCenter and, if in a site with NSX, register the NSX Manager as well. As mentioned in the opening you can also plug into a small subset of popular physical networking equipment such as Cisco, Arista, DELL, Brocade and Juniper.
Once the vCenter has been connected and verified you then have the option to select the vDS and PortGroups you want to have monitored. This enabled Netflow (IPFIX) across all PortGroups selected…it does these changes live so be wary of any possible breaks in vDS traffic flow just in case.
Due to a rather serious PSOD bug in previous version of ESXi when Netflow is enabled, the configurator blocks any host that doesn’t meet the minimum ESXi builds as shown below.
Below is the minimum requirements for Network Insight to be configured and start collecting and analyzing.
- vCenter 5.5 or above
- ESXi 5,5, update 2 (build 2068190) and above
- ESXi 6.0, update 1b (3380124) and above
- NSX for vSphere 6.1 or greater
- Netflow enabled on vDS
Reading through the FAQ, you get to learn about IPFIX and how it’s used with the vDS to collect network traffic data…it’s worth spending some time going through the FAQ however I’ve pulled an overview on how it all works below.
IPFIX is an IETF protocol for exporting flow information. A flow is defined as a set of packets transmitted in a specific timeslot, and sharing 5-tuple values – source IP address, source port, destination IP address, destination port, and protocol. The flow information may include properties such as timestamps, packets/bytes count, Input/output interfaces, TCP Flags, VXLAN Id, Encapsulated flow information and so on.
Network Insight uses VMware VDS IPFIX to collect network traffic data. Every session has two paths. For example: Session A↔C has A→C packets and C→A packets. To analyze the complete information of any session, IPFIX data about packets in both the directions is required. Refer following diagram where VM-A is connected to DVPG-A and is talking to VM-C. Here DVPG-A will only provide data about the C→A packets, and DVPG-Uplink will provide data about A→C packets. To get the complete information of A’s traffic, Ipfix should be enabled on DVPG-A, DVPG-uplink
That wraps up this post…I’ll be looking at doing a followup post that looks at the Network Insight user interface and what information about network traffic, flows and routing can be viewed and analysed as well as taking a look at the surprisingly good VM, Host and Cluster level metrics