TL;DR
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Cloudflare says changes to its DNS build pipeline cut peak memory use by 85% and halved the time from a DNS change to the resulting data being written to Quicksilver. The company attributes the gains to a planner that selects the smallest build it can prove is safe, plus a scheduler that combines work for the same zone. The report does not give absolute delay times or independent measurements.
Cloudflare says it has halved the time between a DNS change and the pipeline writing the resulting data to Quicksilver, while cutting peak memory use by 85%. The company attributes the gains to targeted builds that avoid rebuilding an entire DNS zone when a smaller update can be proven safe, alongside a scheduler that combines work for the same zone.
Cloudflare’s DNS Zone Builder turns changes stored in databases into data its authoritative DNS servers can use. The builder processes events from the DNS Kafka Aggregator, reads the required zone information, converts it into keys, and compares those keys with the existing data in Quicksilver, Cloudflare’s distributed key-value store. Quicksilver then propagates the changes across the company’s network.
The change addresses a scaling problem with full builds. Even when a full build writes only changed keys, it still has to read and process every record in a zone. Cloudflare says some zones now exceed 20 million records, making whole-zone processing costly in memory and work. Its new planner selects a smaller build for eligible changes; its scheduler combines work for the same zone. Cloudflare says zone-setting changes no longer trigger full rebuilds for zones with tens of millions of records.
The company reports an 85% reduction in peak memory and a halving of the time from a DNS change to its data being written to Quicksilver. It says the reduced memory pressure also created enough headroom to shorten waits the pipeline uses for resilience. The post does not provide absolute delay figures, test conditions, or a separate measurement of how much each change contributed.
Faster Updates for Large DNS Zones
DNS changes need to reach the systems that answer queries, and processing them can become more demanding as zones grow. Avoiding unnecessary full builds may let Cloudflare handle large zones with less memory while reducing the time its pipeline spends before writing updates to Quicksilver. That is relevant to customers whose DNS zones contain many records, as well as to the capacity available for other pipeline work.
The reported speed-up is specifically about the interval until the change is written to Quicksilver. Cloudflare describes Quicksilver as propagating data across its network, but the post does not quantify end-to-end time until every location has received a change. The result should not be read as a measurement of all DNS propagation or of how quickly every user worldwide will see an update.
Popular right now
Dreame G10 Pro Wet and Dry Vacuum Mop
Ninja Flip Toaster Oven & Air Fryer, Large Capacity, Stainless, SP151
Vitamix 5200 Professional-Grade Blender, 64 oz. Container, Black
As an affiliate, we earn on qualifying purchases.
Why Whole-Zone Builds Strained
Cloudflare’s earlier account of its DNS propagation pipeline, published in 2022, said the Zone Builder received about 250 DNS record changes per second. In the new report, the company says the current rate is about 700, nearly three times as many, plus roughly 100 other events, including zone-setting changes. Cloudflare says the event count itself was not the main problem; the amount of work required for some events, and how that work grew with zone size, was.
Cloudflare had already introduced per-record builds for record changes. Its 2022 post said those were 150 times faster on average than full builds, and 4,250 times faster on a test zone with one million records. The next challenge was targeting zone-setting changes. Unlike a record event, a setting event may come from different products, may not explain exactly what changed, or may affect data shared by multiple zones. The new planner is designed to choose the smallest build that the company can establish is safe.
That safety requirement shapes the system’s limits. Cloudflare says events now include changed zone fields and settings, with previous and new values when available. If an event is ambiguous, missing details, or could affect records anywhere in the zone, the planner still uses a full build. The company says timestamp-only updates also continue to trigger full builds when it cannot identify whether the update reflects a meaningful change.
“The problem was the kind of work some of those events needed, and how that work grows with the size of a zone.”
— Cloudflare, in its DNS build report
Limits of the Reported Results
The source provides percentage changes but not the absolute time saved, the period over which memory use was measured, or details of the tests and workloads behind the figures. It also does not report independent verification or break down how much of the improvement came from the planner, scheduler, or shorter resilience waits.
Cloudflare describes the measure as time from a DNS change until its data is written to Quicksilver. The report does not specify a corresponding end-to-end propagation time across every point in the network, nor does it quantify customer-visible effects. The company also says some ambiguous events still require full builds; it does not say how frequently those events occur or how much work they account for.
Further Targeting of Zone Events
Cloudflare’s report describes an ongoing effort to identify what different events change and map them to the smallest safe build. The company says timestamp-only updates remain a reason for full builds when their source or purpose is unclear. Identifying those callers and determining whether their updates affect DNS output could allow more events to be handled selectively, though the post does not give a schedule for that work.
The next useful measures would include absolute build-delay figures, results across different zone sizes, and data showing how often the system falls back to full builds. Cloudflare has not announced a date for further results in the supplied report. For now, the published figures describe the company’s reported pipeline changes and their measured memory and timing effects.
Key Questions
What did Cloudflare change?
It introduced a targeted build planner to select smaller, safe updates for eligible DNS events, as well as a scheduler that combines work for the same zone.
What does the 85% figure measure?
Cloudflare says the changes cut peak memory use by 85%. The supplied report does not provide absolute memory figures or details of the measurement period.
Does halving the delay mean DNS changes reach every user twice as fast?
Not necessarily. The reported interval ends when the changed data is written to Quicksilver. Cloudflare does not give an end-to-end measurement for when every network location or DNS user receives the update.
Why do some changes still trigger a full build?
Cloudflare says incomplete or ambiguous events may affect records anywhere in a zone. In those cases, a full build is retained to reduce the risk of stale or incorrect DNS data.
How much time does the change save in absolute terms?
The report says the time from a DNS change to its data being written to Quicksilver was halved, but it does not provide the original or new delay in seconds or minutes.
Source: blog.cloudflare.com
