Advantages of using a SaaS cloud monitoring system
- "Hello, the bulb I bought you the other day doesn't work" - says the client. - Oh, how strange, the bulb in our office works perfectly" - replies the technical service.
We find this situation absurd, don't we? Let's replace the bulb with "web service" and it won't be so strange anymore. In my professional life I have suffered this situation in my flesh once and it was one of the reasons that led me to place much of the monitoring in cloud services in my previous company. Customers would call us and tell us that they couldn't access the service and, as with the light bulb, we were able to access it from our office, so we indicated to the customer that it was likely to be a problem with their ISP. A person on my team came up with the idea of trying the mobile phone, which was connected to another network... and was able to reproduce the problem. So that mobile phone became our improvised probe to check the origin of the problem and solve it. Imagine the customer's face: he could not access the service and we also indicated that it was probably his fault, when from the minute one was a routing problem of our provider.
This first point is perhaps the most obvious advantage of having a monitoring system external to our organization and our network. For a similar reason and because external providers also fail, it is best to adopt a hybrid solution, in which we are able to have external and internal probes (per site or network segment) that report in real time to our centralized cloud system. But there are more advantages:
Like any SaaS (Software-as-a-Service) service in the cloud, using it avoids the need and cost of keeping it in our own datacenter, which saves us hardware costs.
The recommendation is to have the monitoring system in a machine independent of the monitored systems since, for example, if we suffer problems in the storage of our virtualization system, the monitoring will also suffer and it is possible that it prevents us from being warned of this situation. As a note, this also applies to not hosting the monitoring system in the same cloud where we deploy our infrastructure.
In addition, and this is a constant, monitoring needs grow with the use of this and with the growth of the company itself and its services, so either we buy a hardware that we will barely take advantage of at first or we change as needed. In the cloud, this problem is solved by definition. And all this multiplies if we need high availability...
Naturally, other associated costs appear such as electricity consumption, equipment and cybersecurity personnel costs, space in our racks, etc., but perhaps the most important is maintenance and personnel costs: we need internal or external systems personnel who can implement and maintain the monitoring system. These personnel also need specific training in issues such as: maintenance of high load databases, high availability, parameterization of the software used, etc., and must also be up to date with these technologies to perform updates.
In many companies these profiles are not available and they are subcontracted to third parties who have to be given access to our systems for remote actions, they have to adapt and know our internal policies, etc. So, the result is usually not agile, a key factor if we do not want our monitoring system to become obsolete in the face of the rapid advance of new technologies such as containers, the cloud, etc.
Way much cheaper
If the company does have these profiles, in addition to its cost, there is another cost that is usually much higher, although difficult to measure, the opportunity cost: How much does it cost to have 2 engineers, 2 months building a monitoring system and spending X hours a month to maintain it while we could have them doing other projects that generate income for the company? Is the profile of a monitoring engineer useful for other developments or activities that generate income?
In addition, there is a terrible reality: an internally developed system normally means that it will be open to "change requests" of all kinds from internal customers, which will only increase the costs indicated above. This point is especially serious if our company is not even involved in the world of software development (for example, if we are a telecommunications operator). The difficulty of maintaining a system that gains complexity every day, change policies, deployments, documentation, training, etc. can bring out many costs that do not exist to outsource.
Faced with this, a SaaS provider in the cloud has optimized its operations to make massive, fast and tested updates, bearing in mind that its competitiveness as a service is based on its functionalities, as well as having the latest security patches immediately. In addition, we will benefit from the fact that your entire client portfolio is using the same software as us, so the detection of bugs, requests for functionalities and problems and their resolution will be much more agile and, on many occasions, transparent to us.
It is faster
Since SaaS services normally work with a pay-per-use system, the interest in making it very easy to start monitoring easily and quickly is enormous, so they often provide, as in the case of Muutech:
Optimized templates for the most common systems
Good practice oriented
Additional support and configuration service
This has the additional advantage, in addition to a quick implementation, the possibility of being able to test different systems easily and quickly before deciding (also, as in our case, most have an online demo).
Higher availability by design
If we have multiple sites, the fact that the system is out of all of them, eliminates the dependence on the connectivity of one of them or the need to implement high availability strategies between multiple CPDs with systems that, such as monitoring, consume bandwidth and storage. In addition, the provider will be responsible for monitoring their own systems and the security of them, so we do not have to have an additional service to monitor our monitoring system.
Having distributed monitoring using proxies in our headquarters and a central server in the cloud also allows us to establish more intelligent alarms combining data on delays or availability from the central server and from our location; for example, we detect that we have a delay problem connecting to our ERP in the cloud ... the problem is the connection from the office or provider? If we have a probe in our office and another in the cloud server, we can know. This is especially relevant if we have in our infrastructure geographic dispersion, where we also need to avoid delays that affect the efficiency of monitoring systems.
In addition, by applying a hybrid architecture, we have some additional advantages:
We can collect monitoring information in the cloud for computers that do not have a direct connection.
If we are monitoring infrastructure and services in the cloud, this monitoring will not consume bandwidth from our on-premise infrastructure.
We can send data from external systems to our VPN (remote sensors, partner information, data from other SaaS products, etc.).
Other SaaS tools we use can drink data from our monitoring system.
Allows us to publish to our clients information about the state of our services in a direct way.
No need to be an expert in monitoring (or hire them)
There was a time when it was not uncommon for companies to set up their own DNS or email servers on-premise, forcing the company to have (and maintain) people on their computers with knowledge about these technologies, attending conferences on DNSSEC, anti-spam technologies, etc. (surely the opportunity cost we talked about earlier has come to mind). Nowadays it seems unthinkable not to have a critical service like email with cloud services like Google or Outlook and much less to maintain your own DNS system.
The cloud is an economic way to buy knowledge and dedication to specific technologies; taking it to the extreme and trying not to offend anyone, probably Google knows how to manage better mail systems and DNS in high availability than your team: they will be able to get faster maintenance and less downtime, solve any problem sooner and visualize and collect the needs and new features for the future. And the same thing happens with monitoring.
Possibility of access from anywhere in the world, at any time
I receive an alarm indicating that my VPN system is inaccessible and I want to send the report to the technical team of the company that takes it to be resolved. I'm in a hotel and I'm going to connect to the monitoring system of my facilities, wait, how do I connect? A SaaS solution allows, if we want, access from anywhere in the world with the Internet without the need to open ports to our internal network, and so on. Something very useful if, for example, we want to give access to a provider of a particular service to see the status of systems and their history, but not necessarily have remote access to them if we only assist in solving problems.
Ready for the future
In the world of monitoring, a series of changes related to the use of artificial intelligence technologies such as machine learning for the detection of anomalies, prediction of capacity needs, etc. are approaching. The computational cost of many of these activities is high and is often executed discontinuously (hourly, daily, etc.): the cost of doing this with your own hardware is obviously much higher than doing it in the cloud, so for these future functionalities the difference between the cost on-cloud versus on-premise will be much more pronounced. Therefore, why not have it already in the cloud to adopt these improvements before anyone else... avoiding the cost that the future and inevitable migration will bring us?
Finally, we must mention some disadvantages, in most cases bearable or salvageable:
There may come a point where we have a series of special requirements that are not met by any SaaS tool, but here the problem is usually not that the software is in the cloud, but that the software itself does not meet the requirements we need. This has happened in companies such as Netflix, Twitter or Uber, which also have the technical and economic capacity to build a new monitoring system adapted to their specific needs. In fact, some of these companies maintain hybrid systems, taking advantage of what already worked and focusing their development on those particularities not covered by these software. This is equivalent to Salesforce type SaaS solutions, where what is important is the modularity and flexibility of the platform offered by the provider. In the case of Muutech, for example, we work with software that allows extensive flexibility and customization such as Zabbix and Grafana, in addition to offering additional services to meet special needs without touching the core of these tools.
Availability of Internet connection. Nowadays, if there is no Internet access in your offices, the problem is much more serious than remaining without monitoring, with the aggravating factor that if your system is in the office and does not have Internet access, it will not be able to warn you of this problem.
Dependence on the availability of the provider. Obviously, the provider must offer some service SLAs and still Amazon or Azure can fall, but surely the availability will be greater than on-premise and when there are problems you will not have to fix them yourself....
We hope to have shed some light on the decision to install a monitoring system in the cloud versus on-premise. We prefer, for all these reasons, that you use our cloud solution, but we have the option to install it on your datacenter if we haven't convinced you. Good monitoring!