Improved Alerting with Atlas Streaming Eval | by Netflix Expertise Weblog | Apr, 2023

Ruchir Jha, Brian Harrington, Yingwu Zhao

TL;DR

• Streaming alert analysis scales significantly better than the normal method of polling time-series databases.
• It permits us to beat excessive dimensionality/cardinality limitations of the time-series database.
• It opens doorways to assist extra thrilling use-cases.

Engineers need their alerting system to be realtime, dependable, and actionable. Whereas actionability is subjective and should fluctuate by use-case, reliability is non-negotiable. In different phrases, false positives are dangerous however false negatives are absolutely the worst!

A couple of years in the past, we have been paged by our SRE crew because of our Metrics Alerting System falling behind — crucial software well being alerts reached engineers 45 minutes late! As we investigated the alerting delay, we discovered that the variety of configured alerts had just lately elevated dramatically, by 5 occasions! The alerting system queried Atlas, our time sequence database on a cron for every configured alert question, and was seeing an elevated throttle charge and extreme retries with backoffs. This, in flip, elevated the time between two consecutive checks for an alert, inflicting a world slowdown for all alerts. On additional investigation, we found that one consumer had programmatically created tens of 1000’s of recent alerts. This consumer represented a platform crew at Netflix, and their aim was to construct alerting automation for his or her customers.

Whereas we have been in a position to put out the instant hearth by disabling the newly created alerts, this incident raised some crucial considerations across the scalability of our alerting system. We additionally heard from different platform groups at Netflix who wished to construct comparable automation for his or her customers who, given our state on the time, wouldn’t have been in a position to take action with out impacting Imply Time To Detect (MTTD) for all others. Relatively, we have been taking a look at an order of magnitude improve within the variety of alert queries simply over the subsequent 6 months!

Since querying Atlas was the bottleneck, our first intuition was to scale it as much as meet the elevated alert question demand; nonetheless, we quickly realized that may improve Atlas price prohibitively. Atlas is an in-memory time-series database that ingests a number of billions of time-series per day and retains the final two weeks of information. It’s already one of many largest providers at Netflix each in measurement and price. Whereas Atlas is architected round compute & storage separation, and we may theoretically simply scale the question layer to fulfill the elevated question demand, each question, no matter its sort, has a knowledge part that must be pushed all the way down to the storage layer. To serve the rising variety of push down queries, the in-memory storage layer would wish to scale up as effectively, and it turned clear that this could push the already costly storage prices far increased. Furthermore, frequent database optimizations like caching just lately queried knowledge don’t actually work for alerting queries as a result of, usually talking, the final obtained datapoint is required for correctness. Take for instance, this alert question that checks if errors as a % of complete RPS exceeds a threshold of fifty% for 4 out of the final 5 minutes:

identify,errors,:eq,:sum,
identify,rps,:eq,:sum,
:div,
100,:mul,
50,:gt,
5,:rolling-count,4,:gt,

Say if the datapoint obtained for the final time interval results in a constructive analysis for this question, counting on stale/cached knowledge would both improve MTTD or end result within the notion of a false unfavourable, at the very least till the lacking knowledge is fetched and evaluated. It turned clear to us that we would have liked to unravel the scalability drawback with a essentially completely different method. Therefore, we began down the trail of alert analysis by way of real-time streaming metrics.

Excessive Degree Structure

The concept, at a excessive degree, was to keep away from the necessity to question the Atlas database nearly solely and transition most alert queries to streaming analysis.

Alert queries are submitted both by way of our Alerting UI or by API shoppers, that are then saved to a customized config database that helps streaming config updates (full snapshot + replace notifications). The Alerting Service receives these config updates and hashes each new or up to date alert question for analysis to one in all its nodes by leveraging Edda Slots. The node chargeable for evaluating a question, begins by breaking it down right into a set of “knowledge expressions” and with them subscribes to an upstream “dealer” service. Knowledge expressions outline what knowledge must be sourced in an effort to consider a question. For the instance question listed above, the information expressions are identify,errors,:eq,:sum and identify,rps,:eq,:sum. The dealer service acts as a subscription supervisor that maps a knowledge expression to a set of subscriptions. As well as, it additionally maintains a Question Index of all lively knowledge expressions which is consulted to discern if an incoming datapoint is of curiosity to an lively subscriber. The internals listed here are outdoors the scope of this weblog publish.

Subsequent, the Alerting service (by way of the atlas-eval library) maps the obtained knowledge factors for a knowledge expression to the alert question that wants them. For alert queries that resolve to a couple of knowledge expression, we align the incoming knowledge factors for every a kind of knowledge expressions on the identical time boundary earlier than emitting the gathered values to the ultimate eval step. For the instance above, the ultimate eval step could be chargeable for computing the ratio and sustaining the rolling-count, which is conserving monitor of the variety of intervals through which the ratio crossed the edge as proven under:

The atlas-eval library helps streaming analysis for many if not all Query, Data, Math and Stateful operators supported by Atlas as we speak. Sure operators resembling offset, integral, des aren’t supported on the streaming path.

OK, Outcomes?

In the beginning, we now have efficiently alleviated our preliminary scalability drawback with the polling primarily based structure. At this time, we run 20X the variety of queries we used to run a number of years in the past, with ease and at a fraction of what it will have price to scale up the Atlas storage layer to serve the identical quantity. A number of platform groups at Netflix programmatically generate and keep alerts on behalf of their customers with out having to fret about impacting different customers of the system. We’re in a position to keep robust SLAs round Imply Time To Detect (MTTD) whatever the variety of alerts being evaluated by the system.

Moreover, streaming analysis allowed us to chill out restrictions round excessive cardinality that our customers have been beforehand working into — alert queries that have been rejected by Atlas Backend earlier than because of cardinality constraints are actually getting checked appropriately on the streaming path. As well as, we’re in a position to make use of Atlas Streaming to observe and alert on some very excessive cardinality use-cases, resembling metrics derived from free-form log knowledge.

Lastly, we switched Telltale, our holistic software well being monitoring system, from polling a metrics cache to utilizing realtime Atlas Streaming. The elemental concept behind Telltale is to detect anomalies on SLI metrics (for instance, latency, error charges, and so forth). When such anomalies are detected, Telltale is ready to compute correlations with comparable metrics emitted from both upstream or downstream providers. As well as, it additionally computes correlations between SLI metrics and customized metrics just like the log derived metrics talked about above. This has confirmed to be helpful in the direction of lowering Imply Time to Recuperate (MTTR). For instance, we’re in a position to now correlate elevated error charges with elevated charge of particular exceptions occurring in logs and even level to an exemplar stacktrace, as proven under:

Our logs pipeline fingerprints each log message and attaches a (very excessive cardinality) fingerprint tag to a log occasions counter that’s then emitted to Atlas Streaming. Telltale consumes this metric in a streaming style to establish fingerprints that correlate with anomalies seen in SLI metrics. As soon as an anomaly is discovered, we question the logs backend with the fingerprint hash to acquire the exemplar stacktrace. What’s extra is we are actually in a position to establish correlated anomalies (and exceptions) occurring in providers which may be N hops away from the affected service. A system like Telltale turns into more practical as extra providers are onboarded (and for that matter the complete service graph), as a result of in any other case it turns into troublesome to root trigger the issue, particularly in a microservices-based structure. A couple of years in the past, as famous on this weblog, solely a few hundred providers have been utilizing Telltale; due to Atlas Streaming we now have now managed to onboard 1000’s of different providers at Netflix.

Lastly, we realized that when you take away limits on the variety of monitored queries, and begin supporting a lot increased metric dimensionality/cardinality with out impacting the associated fee/efficiency profile of the system, it opens doorways to many thrilling new prospects. For instance, to make alerts extra actionable, we could now have the ability to compute correlations between SLI anomalies and customized metrics with excessive cardinality dimensions, for instance an alert on elevated HTTP error charges could possibly level to impacted buyer cohorts, by linking to exactly correlated exemplars. This is able to assist builders with reproducibility.

Transitioning to the streaming path has been an extended journey for us. One of many challenges was issue in debugging situations the place the streaming path didn’t agree with what’s returned by querying the Atlas database. That is very true when both the information just isn’t accessible in Atlas or the question just isn’t supported due to (say) cardinality constraints. This is among the causes it has taken us years to get right here. That mentioned, early indicators point out that the streaming paradigm could assist with tackling a cardinal drawback in observability — efficient correlation between the metrics & occasions verticals (logs, and probably traces sooner or later), and we’re excited to discover the alternatives that this presents for Observability typically.