Skip navigation

We’ve written about the Hyper9 acquisition on The WhiteBoard and on Geek Speak, but I’m going to add a couple of thoughts and try my best to indicate where Hyper9 and our other related products are headed.

 

First off, why did we buy Hyper9?  As we’ve said before, we listen to our customers.  What we heard them telling us is that virtualization continues to grow in importance.  Moreover, we heard that they’re facing new management challenges around capacity planning, performance optimization, and configuration tracking.  We have a great engineering team, and they could have built what our customers needed, but it’s pretty clear that the problems they face are both serious and urgent, so we looked around at other companies offering virtualization management solutions.  To be honest, it’s not easy to find companies and products to join SolarWinds because their software has to meet the same standards as our software:  Easy to install, configure, deploy.  The UI has to be simple, yet powerful.  It has to be accessible to the little shops yet scale to the really big shops.  It’s not easy to find products like that because it’s not easy to build products like that. 

 

I can’t speak for everyone at SolarWinds, but one of my personal tests for evaluating other people’s software is whether I’m overcome with envy that someone else built something so good and that I didn’t get to be part of it.  That’s how I felt when I saw Hyper9’s product.  It’s pretty awesome.  Why didn’t I do that?  Sigh. My personality disorders aside, the quality of the product really came as no surprise because the product team is actually here in Austin.  Consequently, many of us already knew the folks on the Hyper9 team personally or by reputation.  Awesome product, awesome product team.   We bought Hyper9 because they solved the virtualization management problem so well that seemed like a much better option to join forces on this front than to compete with them.

 

If you have a virtual environment, you should check out Hyper9.  It’s a virtual appliance that you can download and drop on your VMware host.  Simple config process, and you’re collecting data.  You’ll find orphaned VMs.  You’ll be able to see how to squeeze more performance and more capacity out of your infrastructure. It’s got an incredible search engine that lets you find out anything you need about your VMs and their performance.   Plus, it’s just cool to play with that much new data. 

 

I do want to answer one of the questions we’ve started to get on thwack and directly to PMs about how Hyper9 relate to our existing Profiler product line, which has some virtualization-related functionality.  Where does each product fit?  Here’s the deal:  The Profiler product line is focused on storage management, but it has some virtualization features around mapping VMs to storage resources, which allows troubleshooting of VM problems such as those caused by multiple VMs contending for the same storage resources.  If you have a problem managing storage or a problem managing how virtualization impacts storage (or vice versa), then Profiler is the right product.  In contrast, Hyper9 is aimed at broad and deep virtualization problems like VM sprawl, performance optimization, and capacity planning—pretty much anything that isn’t storage.  Anyone who wants to get the most out of their virtual infrastructure needs Hyper9.

 
      
  • If you have a virtualization problem that has nothing to do with storage, you need Hyper9
  •    
  • If you have storage problem, but don’t much care about virtualization, you need Profiler
  •    
  • If you have a problem with the impact of virtualization on storage, you need Profiler
  •    
  • If you have a problems with storage, virtualization, and with the impact of virtualization on storage, you need both Hyper9 and Profiler.
 

And that’s how we’re going to invest in these products going forward:  Profiler will continue to focus on storage management, including storage management problems arising from virtual machines.  Hyper9 will continue to focus on all other aspects of virtualization management.  Will we integrate Profiler and Hyper9?  I can’t promise anything, but it wouldn’t be a shocker if we did, would it?

As many of you know, Storage Manager (Profiler) gets a plethora of data, maybe even too much - but many of you ask about how to set thresholds and alerts so you can be notified when something is amiss.  In Profiler, getting  alerts involves three steps:

  1. Building a rule, which includes a threshold on the metric of interest
  2. Assign it to a policy (ie, the set of resources you want to monitor) and push it out
  3. Setting the Notification to alert you via email when the trap is received.

For the threshold, lets focus on performance metrics right now -  although you can do storage and asset change thresholds as well.

Go to Settings > All Rules > Add New Rule.  From the list of  choices, choose Threshold Rule.  You should see the following screen:

Some quick definitions:

  • Section - basically the scope of resources this rule would apply to  (Ex: NetApp)
  • Category - the types of metrics applicable to that section (Ex: LUN Performance)
  • Instances (if applicable) - the instances of the metric we are monitoring (All instances)
  • Condition - the threshold on the metric. (Average Latency (ms) > 20)
  • Duration - how long the condition has to be met before the threshold is triggered (0 Min)
  • Choose Action - choose one action (Send Trap)

 

So what we are telling Profiler in this example is to send us a trap  whenever any instance of a NetApp LUN has average latency greater than  20ms. Before moving to the next step, a couple of cool things:

  • When you set Profiler to Any Instances, new objects are covered  automatically.  If you create a new LUN, Profiler will automatically  apply the rule to that instance.
  • You can pick one or more instance - so you can get very particular if you need to.
  • The duration allows you to filter out noise, so you don't get alerted on every little spike.

So, you have your rule, now you have to apply it.  In Profiler, you  do that via policies - which are just a collection of resources of the  same type that you configure at the same time.  Every resource type has a  Default Policy, and that is the one we will use today.

Go to Settings > Policies and click the edit icon for Default NetApp Filer Policy (let us stick with NetApp for this example)

Click Rules and you will see a list of rules that are available to be  assigned, or already assigned to the policy.  Note there are default  rules already assigned to identify problems for you.  To assign a rule  to the policy, click the rule and press the down arrow, and then press  the Save button.

Now the rules is assigned to the policy - but - make sure you press  the Push button to update the configuration on the agents monitoring the  NetApp Filers.

So now, if a condition were met, the agent would send a trap to the  Profiler server and you would see the trap in the Event Monitor. You  could then manage the event

However, if you want to receive an email for that event, you need to  turn on notifications.  Notifications are turned on per user, so go to  Settings > Users and click the edit icon for your login.  If you have  defined an email address, you will see a "Notifications" section.   Click the Add button in this section.

Now you can add a notification for on resource or a group of  resources.  Choose "Groups" and then choose "All Devices".  You can then  pick the trap severity you want to be notified, and one or more email  addresses to send the email to.

Whew! That was a few too many steps (hint, we will make this better in the future) - but now I can safely sleep knowing that I will be notified if I have a problem.

As a bonus, I'll throw in a few notes about managing the events on the Event Monitor:

  • Events that occur over and over again for the same object only notify you on the first occurrence, but maintain a count thereafter (hence the count column in the Event Monitor)
  • You can acknowledge and clear traps thru the event monitor
  • In the Setting > Server Setup > Server, you can turn the automatic clearing of events after a certain amount of time.

Thanks for listening - and as always, if you have thoughts or feedback, we would love to hear it.

Date/Time: Thursday, Feb 10, 2011 11:00 AM- 12:00 PM CST

Register: https://www1.gotomeeting.com/register/707934872

Please join us on Thursday, February 10th at 11:00 AM CST for an exclusive training session on network configuration management.   

During this webcast we’ll demonstrate best practices for network configuration management  – managing the configurations of routers, switches, firewalls and other network infrastructure devices. We’ll also demonstrate the top 5 network configuration management tasks using the SolarWinds Network Configuration Manager (NCM).

Some of what we’ll discuss and demonstrate will include:

·        Backing up the configurations of all of your routers, switches, and firewalls

·        Setting up automated configuration management jobs

·        Configuring and using real-time config change detection

·        Managing configurations against policies for compliance

This training session is free for all attendees.

We often get feedback from users on what they would like to see in the products. Sometimes the requests fit into our existing solutions, and sometimes, we decide to focus more extensively on the problem by creating a new product.

 

We are actively working on two new products that our users have requested.

 
      
  • A product to help monitor synthetic web transactions. If you are interested in the web transaction product, contact Christine Bentsen (thwack ID christineb) and she will give you more information.
  •    
  • A product that gives you more detailed information about where devices are plugged in to your network. If you are interested in this, read on!
 

For the user tracking product we are helping to address a long time request from many of our users, the ability to quickly find where a certain device is connected in the network. For example, search for the MAC address 0123.45.67.89ab, the results will include what switch and port that MAC address is connected to. See the screenshots below to get an idea of what this will look like.

 

_search_results3

 

_search_results4

 

 

 

Here is where we need your help. Did you notice the difference between the two screenshots? The first image incudes status directly in the icon, the second image indicates the status by using the ball icon. Which do you prefer? Just post below and we will gather your comments together and implement the solution that works the best!

 

Of course, this is not all the product will do. Stay tuned to see more screenshots and information about what we are working on for this. Also, we will be demonstrating some basic functionality for this new product at Cisco Live! in London (booth G4). Stop by the booth and let us know what you think in person! If you are really excited about this, please take this survey and we will include you in Beta opportunities and further chances to provide direct feedback with our developers and UI team!

We’re running a “Getting Started with SolarWinds NPM” training session on  Thursday Jan 27, 2011 at 11:00 AM CST.  To participate, you need to register here:  https://www1.gotomeeting.com/register/989188488

  • During this 60 minute training session we’ll cover: 
  • Optimizing hardware configurations for Orion
  • Understanding Orion’s architecture
  • Installation, discovery, and base configuration
  • Leveraging the information that Orion provides

This class is most beneficial to current users of Orion NPM and will be hosted by Josh Stephens, our Head Geek, and Brandon Shopp, Sr. Product Manager for Orion NPM.

Over the past couple of months the Product Management team has been walking through all the new features in 10.1 and another one of the key features I would like to touch on a bit is our addition of native out of the box support for the Cisco Unified Computing System or UCS for short.

 

So what is this UCS thing?

 

When Cisco first announced the UCS, the industry positioned this as Cisco’s entry into the server market.  The UCS is designed to allow customers to build programmable data center infrastructure optimized for virtualized resource, as well as memory-intensive bare-metal workloads.. It is also intended to manage all physical and logical UCS elements—servers, I/O, VMs, etc-- as a unified environment and support applications and services from leading vendors, including Microsoft, EMC, VMware, Red Hat and Novell.

 

I borrowed the below graphic from the cisco.com website to help illustrate the pieces that go into the UCS chassis.

 

image

 

Why did we add support for the Cisco UCS in NPM 10.1?  Two primary reasons:

 
      
  1. Based on the increased support demand for the Cisco Nexus product lines, we felt the next natural step was to also support the UCS.  As you can see here in NPM 10.1 UCS Monitoring - Adding Blades, SNMP Possible? thread, there has already been quite a bit of discussion on thwack about UCS support in Orion.
  2.    
  3. It is my opinion that the line that used to exist between the server, application and networking teams are blurring together.  No longer does the network end at the wall jack or the switch port.    
    Ex. 1 – VMWare vSwitch and vDS or virtual distributed switch resides on the ESX server that contains a software based switch.  Who owns and manages this?     
    Ex. 2 – the Cisco UCS includes a fabric interconnect with physical Ethernet ports and fibre channel ports, which distributes converged traffic through “extended line cards” to server blades with virtual ports running VN-Link technology.
 

As these complexities of running your IT infrastructure increase, so must your IT Management solutions.

 

I can already hear your next question, “Can’t I just use the UCS Manager software.  Why do I need Orion?”  Orion is not meant to replace UCS Manager.  UCS Manager is great for advanced provisioning and management of the UCS itself, but does not extend to other types of infrastructure.  Orion NPM provides the following advantages:

 
      
  • single pane of glass to monitor your wired and wireless infrastructure, as well as your servers (both physical and virtual) and applications
  •    
  • UCS Manager only retains a week’s worth of historical data, whereas Orion can retain it for as long as required
  •    
  • Orion is pulling UCS information directly from the UCS API as well as SNMP, which allows Orion to report and alert on UCS events
 

Let’s walk through adding a UCS chassis to Orion and what it looks like after we start monitoring it.  You will notice in adding a new node there is a new check box dialog for UCS Manager credentials.  Also, you will want to enter in the primary fabric interconnect IP Address of the chassis as this is what we will use to monitor the UCS.

 

image

 

Once added and we start polling the key new data we are gathering as shown in the screenshot below:

 
      
  1. Fabric Interconnects status
  2.    
  3. Chassis and blade status including fan and power supply (PSU) status
  4.    
  5. UCS Events (polled from the API)
  6.    
  7. Fibre Channel information including Connectivity Unit Status, VSAN’s and WWN Port information
 

UCS-Orion-Node-Detailpage

 

We are also gathering port/interface data for all the virtual interfaces, fabric interconnect ports etc. as seen below.

 

 

 

Screen shot 2010-11-22 at 10.02.29 AM

 

 

 

 

 

Tell us what you think.  Are you using Cisco UCS?  What else would you like to see in Orion as it pertains to monitoring Cisco UCS?

 

Also if you are going to Cisco Live in London, come on by the SolarWinds booth to learn more about UCS and Orion NPM 10.1 and get some cool free geeky stuff.  We will be in booth # G4.

Filter Blog

By date: By tag: