Key Pillars of a Modern Analytics & Monitoring Strategy for Hybrid Cloud
>> Good morning, everyone. My name is Sudip Datta. I head up product management for Infrastructure Management and Analytics at CA Technologies. Today I am going to talk about the key pillars for modern analytics and monitoring for hybrid cloud. So before we get started, let's set the context. Let's take a stock of where we are today. Today in terms of digital business, software is driving business. Software is the backbone, is the driving force for most of the business services. Whether you are a financial institution or a hospitality service or a health care service or even a restaurant service pizza, you are front-ended by software. And therefore the user experience is of paramount importance. Just to give you some factoids. Eighty-three percent of U.S. consumers say that the brand that, the frontal software portal is more important than the product itself. And the companies are reciprocating by putting a lot of emphasis on user experience, as you see in the second factoid. The third factoid, it's even more interesting that 53% of the users of a mobile app actually abandon the app if the app doesn't load within a specified time. So we all understand now the importance of user experience in today's business. So what's happening to the infrastructure underneath that's hosting these applications? The infrastructure itself is evolving, right? How? First of all, as we all know there is a huge movement, a huge shift towards cloud. Customers are adopting cloud for reasons of economy, agility and efficiency. And whether you are running on cloud or on prem, the architecture itself is getting more and more dynamic. On the server side we hear about server-less computing. More and more enterprises are adopting containers, could be Dockers or other containers. And on the networking side we see an adoption of software-defined networking. The logical overlay on top of the physical underlay is abstracting the network. While we see a huge shift, a movement towards cloud, it is also true that customers are also retaining some of their assets on prem, and that's why we talk about hybrid cloud. Hybrid cloud is a reality, and it's going to be a reality for the foreseeable future. Take for example a bank that has its systems of engagement on public cloud, and systems of records on prem deeply nested within their DNC. So the transaction, the end-to-end transaction has to traverse multiple clouds. Similarly we talk to customers who run their production tier one application on prem, while tier two and tier three desktop applications run on public cloud. So that's the reality. Multi-cloud dynamic environment is a reality of today. While that's a reality, they pose a serious challenge for IT operations. What are the challenges? Because of multiple clouds, because of assets spanning multiple data centers, multiple clouds, there are blind spots getting created. IT ops is often blindsided on things that are happening on the other side of the firewall. And as a result what's happening is they're late to react, and often they react to problems much later than their customers find it, and that's an embarrassment. The other thing that's happening is because of the dynamic nature of the cloud, things are ephemeral, things are dynamic, things come and go, assets come and go, IT ops is often in the business of keeping pace with these changes. They are reacting to these changes. They are trying to keep pace with these changes, and silo'd tools are not the way to go. They are trying to keep up with these changes, but they are failing in doing so. And as a result we see poor user experience, low productivity, capacity problems and delayed time to market. Now what's the solution? What is the solution to all these problems? So what we are recommending is a four-pronged solution, what we represent as four pillars. The first pillar is about dynamic policy-based configuration and discovery. The second one is unification of the monitoring and analytics. The third one is contextual intelligence, and the fourth one is integration and collaboration. Let's go through them one by one. First of all, in terms of dynamic policy-based configuration, why is it important? I was talking to a VP of IT last week, and he commented that the time to deploy the monitoring for an application is longer than the time to deploy the application itself, and that's a shame. That's a real shame because in today's world application needs to be monitored straight out of the box. This is compounded by the fact that once you deploy the application, the application today is dynamic, as I said, the cloud assets are dynamic. The topology changes, and monitoring tools need to keep pace with that changing topology. So we need automated discovery. We need API driven discovery, and we need policy-based monitoring for large scale standardization. And last but not the least, the policies need to be based on dynamic baselines. The age, the era of static thresholds is long over because static thresholds lead to false alerts, resulting in higher opics for IT, and IT personnel absolutely, absolutely want to move away from it. Unified monitoring and analytics. This morning I stumbled upon a Lincoln white paper which said 20 tools you need for your hybrid monitoring, and I was absolutely dumbfounded. Twenty tools? I mean, that's a conversation non-starter. So how do we rationalize the tools, minimize the silos, and bring them under single pane of glass, or at least minimal panes for glass for monitoring? So IT admins can have a coherent view of servers, storage, network and applications through a single pane of glass? And why is that important? It's important because it results in lesser blame game. Because of silo'd tools what happens is admins are often fighting with each other, blaming each other. Server admins think that it's a storage problem. The storage admin thinks it's a database problem, and they are pointing to each other, right? So the tools, the management tools should be a point of collaboration, not a point of contention. Talking about blame game, one area that often gets ignored is the area of fault management and monitoring. Why is it important? And I will give a specific example. Let's say you have 100 VMs, and all those VMs become unreachable as a result of router being down. The root cause of the problem therefore are not the VMs, but the router. So instead of generating 101 alarms, the management tool needs to be smart enough to generate one single alarm. And that's why fault management and root cause analysis is of paramount importance. It suppresses unnecessary noise and results in lesser blaming. Contextual intelligence. Now when we talk about the cloud administrator, the cloud admin, the cloud admin in the past were living in the cocoon of their hybrid infrastructure. They were managing the hybrid infrastructure, but in today's world to have an end-to-end visibility of the digital chain, they need to integrate with application performance management tools, APM, as well as what lies underneath, which is the network, so that they have an end-to-end visibility of what's happening in the whole digital chain. But that's not all. They also need what we call is the context of the application. I will give you a specific example. For example, if the server runs out of memory when a lot of end users log into the system, or run out of capacity when a particular marketing promotion is running, then the context really is the business that leads to a saturation in IT. So what you need is to capture all the data, whether they come from logs, whether they come from alarms, capacity events as well as business events, into a single analytics platform and perform analytics on top of it. And then augment it with machine learning and pattern recognition capabilities so that it will not only perform root cause analysis for what happened in the past, but you're also able to anticipate, predict and prevent future problems. The fourth pillar is collaboration and integration. IT ops in today's world doesn't and shouldn't run in a silo. IT ops need to interact with dev ops. Within dev ops developers need to interact with QA. Storage admins need to collaborate with server admins, database admins and various other admins. So the tools need to encourage and provide a platform for collaboration. Similarly IT tools, IT management tools should not run standalone. They need to integrate with other tools. For example, if you want monitoring straight out of the box, the monitoring needs to integrate with provisioning processes. The monitoring downstream needs to integrate with ticketing systems. So integration with other tools, whether third party or custom developed, whatever it is, it's very, very important. Having said that, having laid what the solution should be, what the prescription should be, how is CA Technologies gearing up for it? In CA we have the industry's most comprehensive, the richest portfolio of infrastructure management tools, which is capable of managing all forms of infrastructure, traditional, private cloud, public cloud. Just to give you an example, in private cloud we support the traditional VMs as well as hyper converged infrastructure like Nutanix. We support Docker and other forms of containers. In public cloud we support the monitoring of infrastructure as a service, platform as a service, software as a service. We support all the popular clouds, AWS, Azure, Office 365 on Azure, as well as Salesforce.com. In terms of network, out net ops tools manage the latest and greatest SDN and SD-WAN, the VMware SDN, the open stack SDN, in terms of SD-WAN Cisco, Viptella. If you are a hybrid cloud customer, then you are no longer blindsided on things that are happening on the cloud side because we integrate with tools like Ixia. And once we monitor all these tools, we provide value on top of it. First of all, we monitor not only performance, but also packet, flow, all the net ops attributes. Then on top of that we provide predictive insights and learning. And because of our presence in the application performance management space, we integrate with APM to provide application to infrastructure correlation. Finally our monitoring is integrally linked with our operational intelligence platform. So in CA we have an operational intelligence platform built around CA Jarvis technology, which is based on open source technology, Elastic Logstash and Kibana, supplemented by Hadoop and Spark. And what we are doing is we are ingesting data from our monitoring tools into this data lake to provide value added insights and intelligence. When we talk about big data we talk about the three Vs, the variety, the volume and the velocity of data. But there is a fourth V that we often ignore. That's the veracity of the data, the truthfulness of data. CA being a leader in monitoring space, we have been in the business of collecting and monitoring data for ages, and what we are doing is we are ingesting these data into the platform and provided value added analytics on top of it. If you can read the slide, it's also an open framework we have the APIs from for ingesting data from third-party sources as well. For example, if you have your business data, your business sentiment data, and if you want to correlate that with IT metrics, how your IT is keeping up with your business cycles, you can do that as well. Now some of the applications that we are building, and this product is in beta as you see, are correlation between the various events, IT events and business events, network events and server events. Contextual log analytics. The operative word is contextual. There are a plethora of tools in the market that perform log analytics, but log analytics in the context of a problem when you really need it is of paramount importance. Predictive capacity analytics. Again, capacity analytics is not only about trending, right? It's about what if analysis. What will happen to your infrastructure? Or can your infrastructure sustain the pressure if your business grows by 2X, for example? That kind of what if analysis we should be able to do. And finally machine learning, we are working on it. Out of box machine learning algorithm to make sure that problems are not only corrected after the fact, but we can predict problems. We can prevent the problems in future. So for those who may be listening to this might be wondering where do we start? If you are already a CA customer, you are familiar with CA tools, but if you're not, what's the starting point? So I would recommend the starting point is CA Unified Infrastructure Manager, which is the market leading tool for hybrid cloud management. And it's not a hollow claim that we are making, right? It has been testified, it has been blessed by customers and analysts alike. And you can see it was voted the cloud monitoring software of the year 2016 by a third party. And here are some of the customer experiences. NMSP, they were able to achieve 15% productivity improvement as a result of adopting UIM. A healthcare provider, their meantime to repair, MTTR, went down by 40% as a result of UIM. And a telecom provider, they had a faster adoption to cloud as a result of UIM, the reason being UIM gave them for the first time a single pane of glass to manage their on prem and cloud environments, which has been a detriment for them for adopting cloud. And once they were able to achieve that, they were able to switch onto cloud much, much faster. Finally, the infrastructure management capabilities that I talked about is now being delivered as a turnkey solution, as a SAS solution, which we call digital experience insights. And I strongly, strongly encourage you to try UIM via CA digital experience insights, and here is the URL. You can go and sign up for the trial. With that, thank you.
SUMMARY :
And on the networking side we see an adoption of
SENTIMENT ANALYSIS :
ENTITIES
Entity | Category | Confidence |
---|---|---|
101 alarms | QUANTITY | 0.99+ |
100 VMs | QUANTITY | 0.99+ |
53% | QUANTITY | 0.99+ |
20 tools | QUANTITY | 0.99+ |
Twenty tools | QUANTITY | 0.99+ |
15% | QUANTITY | 0.99+ |
Eighty-three percent | QUANTITY | 0.99+ |
second factoid | QUANTITY | 0.99+ |
fourth V | QUANTITY | 0.99+ |
40% | QUANTITY | 0.99+ |
CA | LOCATION | 0.99+ |
third factoid | QUANTITY | 0.99+ |
fourth pillar | QUANTITY | 0.99+ |
first pillar | QUANTITY | 0.99+ |
2X | QUANTITY | 0.99+ |
last week | DATE | 0.99+ |
CA Technologies | ORGANIZATION | 0.99+ |
Today | DATE | 0.99+ |
AWS | ORGANIZATION | 0.99+ |
Cisco | ORGANIZATION | 0.99+ |
NMSP | ORGANIZATION | 0.99+ |
four pillars | QUANTITY | 0.98+ |
2016 | DATE | 0.98+ |
third one | QUANTITY | 0.98+ |
first time | QUANTITY | 0.98+ |
Sudip Datta | PERSON | 0.98+ |
fourth one | QUANTITY | 0.98+ |
Hadoop | ORGANIZATION | 0.98+ |
today | DATE | 0.98+ |
First | QUANTITY | 0.97+ |
Office 365 | TITLE | 0.97+ |
one single alarm | QUANTITY | 0.97+ |
second one | QUANTITY | 0.97+ |
Elastic Logstash | ORGANIZATION | 0.96+ |
Azure | TITLE | 0.96+ |
UIM | ORGANIZATION | 0.95+ |
single pane | QUANTITY | 0.95+ |
Lincoln | ORGANIZATION | 0.95+ |
U.S. | LOCATION | 0.95+ |
Kibana | ORGANIZATION | 0.95+ |
This morning | DATE | 0.95+ |
three Vs | QUANTITY | 0.93+ |
one area | QUANTITY | 0.87+ |
one | QUANTITY | 0.86+ |
Viptella | ORGANIZATION | 0.84+ |
VMware | TITLE | 0.82+ |
Nutanix | ORGANIZATION | 0.81+ |
single analytics | QUANTITY | 0.8+ |
Spark | ORGANIZATION | 0.75+ |
four-pronged | QUANTITY | 0.69+ |
Salesforce.com | ORGANIZATION | 0.67+ |
Docker | TITLE | 0.67+ |
tier three | QUANTITY | 0.62+ |
CA | ORGANIZATION | 0.61+ |
Ixia | TITLE | 0.6+ |
tier two | QUANTITY | 0.57+ |
Jarvis | ORGANIZATION | 0.56+ |
APM | ORGANIZATION | 0.54+ |
prem | ORGANIZATION | 0.53+ |
tier one | QUANTITY | 0.53+ |