Confused about the price of an application and server monitoring tool? So were we! Every tool is priced differently and there are a lot of nuances. We’ll walk you through important terminology differences, the pros and cons of different plans, and then discuss the pricing details for Librato, New Relic, SignalFX, Datadog, and of course, Instrumental.
These are simply unique things you’re monitoring. If you monitor server1.cpu and server2.cpu, those are two metrics. If you monitor myapp.page_response.time across 50 cloud servers, that’s just one metric.
Application Monitoring, a.k.a. Custom Metrics
Data on your application, gathered from code you insert into your application. This is how you measure whether your product (and the APIs it uses) are actually working.
Server metrics are basic stats on your actual machines or cloud environments (CPU, disk space, etc).
Measures Apache, MySQL, Mongo, etc.. Everything that’s not application or server monitoring.
The amount of data processed in a time period. Server monitoring data has relatively consistent volume (your servers are always running), while application monitoring volume can vary wildly as your application usage spikes or decreases.
Data Storage & Resolution
An important differentiator to consider when evaluating monitoring prices. Data storage is how long your data is stored at a given resolution. If you want to compare current data to last month, last quarter, or last year, that data needs to be stored. This data retention can be expensive for monitoring providers, so they often limit their storage by dropping data as time passes to make their prices lower.
Comparing Pricing Dimensions
Pricing Based on Host Count: Datadog, New Relic
These platforms charge you for each server you use. 5 database servers + 10 application servers + 5 job servers = 20 servers you have to pay for. This sounds great and simple at first, but it ignores the subtle differences between server types or cloud infrastructure, and removes flexibility. If, for greater redundancy, you wanted to switch from two large AWS servers to four smaller AWS servers, your monitoring costs would double.
And perhaps even more importantly, host-based providers tend to put less emphasis and more pricing complexity on custom application monitoring. Without measuring your application at the code level, you can’t be confident it’s performing as required.
Pricing Based on Volume: SignalFx
SignalFx charges based on the volume of data sent. While reasonable for server monitoring, this can be a significant downside for monitoring high-traffic applications. Applications can often have unpredictable or growing data volumes, whereas server monitoring levels are more constant.
Pricing Based on Metric Count: Instrumental, Librato
Librato and Instrumental both charge per metric. You get exactly what you pay for with both services – you’re only charged for data you want to know. You can also reduce costs by monitoring less on particular hosts.
Pricing Based on Resolution: Librato
In addition to pricing per metric, Librato also charges per resolution of aggregation. Storing a metric at a 5 second resolution costs more than storing it a 60s resolution.
Librato bases their pricing on two dimensions: number of metrics and the resolution of aggregation on those metrics. Librato lets you store a metric at as low as a 5s resolution or up to 300s. If you send metric measurements more frequently than the resolution for that metric, Librato will aggregate the measurements into a single data point.
By default, a metric’s resolution will be defined by how frequently you send data to Librato, but you can also define a specific resolution for a metric (they’ll aggregate any data you send more frequently than that).
Pricing scales up if you want lower resolutions. A 5s resolution costs $0.30/metric/month, whereas a 60s resolution only costs $0.10/metric/month.
Regardless of your initial resolution choice, Librato will automatically aggregate and store your data at 60s after two days, 15m after one week, and 1h resolution after one month.
Server Monitoring: About $5/server/month (a dual-CPU server at a 60s resolution)
Service Monitoring: Varies wildly, but probably $2-$10/mo
Application Monitoring: $30/application/month (300 metrics at a 60s resolution)
Figuring out your needs and estimated costs can be complex with Librato, as can optimizing your spend. A 60s resolution is good enough for almost every scenario (it’s not like you can take action or debug a situation much faster), but don’t forget to set the resolution for all your metrics. You could end up paying for much smaller resolution than you need or use. Fortunately, Librato calculates your cost on an hourly basis, which allows you to adjust your resolution at any time.
New Relic APM pricing is very complex. They charge per CPU core, per GB of RAM, per server. Depending on your servers, theor Essentials package is $25-$100/server/month and the Pro package is $50-$200/server/month. The Essentials package only offers 3 days of general data retention and the Pro package offers just 90 days. Furthermore, the Pro package only includes 8 days of event data. This is significantly shorter data storage than other competitors.
It is important to note that New Relic APM provides more functionality than the metric monitoring other tools in this post include. New Relic APM includes error traces, SQL traces and thread profiling.
New Relic also offers a free Lite plan, but it only includes one day of storage and much fewer features than the paid plans. Hopefully you’re good enough to fix all performance issues very quickly!
Server Monitoring: $50-$200/server/month
Service Monitoring: Included in server monitoring
Application Monitoring: Included in server monitoring
We think New Relic is pretty expensive for a modern cloud infrastructure (frequently more expensive than the servers they are monitoring) and their custom application monitoring capabilities and interface lag other providers. Don’t get sucked in by the free Lite plan and its extremely limited storage. That said, if you do need mobile monitoring and browser monitoring, New Relic might be a good fit for your organization.
Datadog charges $18/server/month, but each server includes 100 custom metrics. Each server provides unlimited integrations and service monitoring. They also offer a free plan with all the features of the paid plans, but only 24 hours of data storage and no alerts.
By default, most of Datadog’s agents aggregate metrics server-side and then send metrics to Datadog every 10 seconds. Datadog stores all data for 13 months at a 1s resolution. That sounds pretty great, but if you send multiple data measurements to their servers within 1s, Datadog only stores the last data point. Datadog gets around this data-dropping limitation by sending all custom metrics to the Datadog agent running on your servers, which aggregates them on your server and sends it along to the Datadog servers. This results in additional infrastructure for you to manage, but minimizes the impact of their limited data reception.
Server Monitoring: $18/server/month
Service Monitoring: Included with server monitoring
Application Monitoring: 100 free metrics for each server (you can $10/month for additional buckets of 200 metrics)
Datadog is a fantastic budget option with a bit more infrastructure work on your end than other tools. The downside is that within both Datadog’s interface and infrastructure, it is highly focused on server monitoring and not application monitoring.
SignalFx charges based on the amount of data points you send them an hour. Specifically, SignalFx charges $25/mo for up to 1,000/data points an hour. Additional data point volume can be purchased at that same rate.
Given the difficulty of estimating how many data points you may wish to send, SignalFx estimates their plans based on the number of hosts monitoring. Each host is estimated to consume 1,000/data points an hour – but this is only an estimate. You will actually be billed based on the data point volume, which may be higher or lower than the estimated 1,000/data points per hour, especially if you implement custom application monitoring with a high-load application.
SignalFx stores data at 1-5s resolution for 8 days, 1 minute for 32 days, 5 minutes for 96 days, and 1 hour for 13 months.
Server Monitoring: $25/mo
Service Monitoring: Included with server monitoring
Application Monitoring: 1,000/data points/hour for $25/mo (the cost will be greatly dependent on the activity level of your application and the amount of pre-SignalFx aggregation you do)
SignalFx’s pricing is complex to estimate, but does provide greater flexibility than host-based pricing models. For users intending to submit many thousands of application data points per minute, you might consider other solutions or using an aggregator such as StatsD.
Instrumental’s pricing is based on what you choose to measure, not arbitrary technical details like volume or hosts. You can submit thousands of data points a second to a metric and it will still cost the same. The price-per-metric varies from $0.50 to $0.25 (and lower for enterprise customers) depending on the plan level.
Instrumental’s metric-based pricing makes it ideal for heavy application monitoring and for cloud-based infrastructure monitoring. Server monitoring can be customized to only send basic metrics for each server or container, thereby greatly reducing cost.
Unlike most other providers, all data is stored at a 60s resolution for at least 13 months. There is no automatic aggregation of data into a larger time resolution, which makes it easy to do detailed historical analysis.
Server Monitoring: $15/mo (decreases dramatically with lower plans or customized monitoring)
Service Monitoring: $15-$30/mo (ditto)
Application Monitoring: $150/application/mo (300 metrics)
We think Instrumental’s plan provides the most value – you should pay to know, not to grow. No counting hosts, no worrying about volume, no counting team members.
There are many great monitoring tools available, but their pricing can vary dramatically between different infrastructure setups and application monitoring levels. Some pricing models greatly restrict your ability to fine-tune your monitoring costs, where some (such as SignalFX and Instrumental) price directly on your usage and give you full control.
At Instrumental, we think our pricing is most directly aligned to the value you receive from Instrumental, and it provides you with the most direct control. You only pay for the metrics you’re actively monitoring and you can adjust these metrics at any time between different applications, microservices, server types, and environments. Take a free 30-day trial and see the flexibility and power for yourself.
This post was last updated in October 2016 and reflects the current state of each product. We will update the post if new information or pricing becomes available. Please contact us if you see any incorrect data. Thanks!