OddEye monitoring system is SaaS platform which provides all in one monitoring and anomaly detection. The main idea behind of OddEye is to have the most advanced monitoring and anomaly detection system,with feature rich dashboards and intuitively guessable configuration parameters.
At some circumstances, it’s hard and sometimes even practically impossible to set up static thresholds for alerting, mostly because of the nature of particular metrics, it may depend on daytime, peak or non peak usage period of particular system.
For example if visitors of platform are from one or close to each other geographical regions, system statuses and load of servers can be significantly different, based on day time,so if at peak hour high CPU usage is expected,but at low hours this is anomaly and vice versa.
OddEye uses machine learning to detect correct thresholds for your systems, which will trigger alerts not only if value of particular metric is higher than expected value, but also if it is lower than it should be. For example if CPU usage at peak hour should be at 80%, but somehow because of network issues, DNS problems or any other reason it is 20%, OddEye will detect this as an anomaly and trigger alert.
Image below demonstrates this kind of situation:
Here we can see two anomalies with unexpectedly higher and lower values for particular time.
Beside of machine learning, OddEye also allows users to configure manual threshold, which will be shown in dashboard as “Special” alerts:
A good example of that can be System Load Average.
Regardless of peak or non peak time, Load Average of Linux systems, that are servings real time clients should be below than amount of cores of CPU. For example if you have Quad Core CPU with Hyperthreading, Linux will see 8 cores, which means that Load Average should be less than 8, otherwise you will have loaded system queue which will cause slowness on serving requests, thus making queue bigger.
Image below illustrated both static and machine learned alerts:
Alerts are first row are triggered by manually defined thresholds, at second row alerts are based onr thresholds defined by machine learning algorithms . As OddEye servers and Clients are highly configurable, you can enable or disable both manually defined, and machine learned alerts as well, as define any threshold, based on your specific case.
Advantages of OddEye
- Pay-As-You-Go (Pay for only what you use )
- No long time commitments ( Use system only when you need it)
- Configurable metrics injections rate.
- At least one year of data retention period.
- No automatic downsampling of data (We will never auto average data so you can see your full data at any point of time)
- Highly configurable alerting and machine learning thresholds with reasonable defaults.
- Intuitive and advanced dashboard query builder.
- Beautiful dashboards.
- Line, Funnel, Gauge, Pie, Table, Single Metric charts.
- Raw Json query builder for advanced users and cases.
- Machine learned and statically defined alerts.
- Always up to date alert levels for machine learning (machine learning never stops.)
- Open Source Agents.
- 45 supported platforms at the moment of writing this paper. (We add more systems almost every week.)
- Open API for building own agents.
- Preconfigured dashboard templates (You just need to clone one and set several parameters.)
- Enterprise support and creation of non standard checks for your particular needs
- Advanced raw json Query builder.
Try OddEye for free. When you register we will add $50 to you balance which you can spend to monitor your servers.
You can use our demo account via following URL and credentials.
URL: https://app.oddeye.co Demo User: email@example.com Demo Pass: 123456