TL;DR
Get the latest gadgets delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
In an interview with The Pragmatic Engineer, Peter Mattis discusses lessons from building storage systems at Google and distributed databases at Cockroach Labs. He also describes how AI tools have brought him back to writing code and argues that domain expertise can help engineers use them to improve their work.
Peter Mattis, co-founder and CTO of Cockroach Labs, discussed distributed database design, Google’s storage systems and AI-assisted software development in an interview published by The Pragmatic Engineer. The episode connects engineering decisions at Gmail and Google’s file-storage systems with the problems of keeping databases fast, reliable and correct as they grow.
Mattis described work spanning several stages of his career: creating the open-source image editor GIMP with college roommate Spencer Kimball, working at Google on Gmail and distributed storage, and later co-founding Cockroach Labs. The interview also covers his recent use of AI coding tools. According to the episode description, Mattis says AI has brought him back to writing code after his work shifted toward management, and that he feels more productive without a drop in quality. That is his assessment of his own work, not a measured result presented in the supplied material.
One example from Google concerns Gmail’s early storage layer. The source says incoming messages were matched to email threads using a search index, while B-trees tracked threads and unread counts. B-trees also recur in the account of Mattis’s later work: he built a B-tree with semantics similar to C++’s std::map after noticing the standard container in memory profiles. The source says the replacement was faster and smaller, attributing the gains to spatial locality and fewer pointers.
The interview also describes Colossus, Google’s successor to the Google File System. The source says Mattis and colleagues used Reed–Solomon erasure coding so the system could store data twice while increasing redundancy, compared with GFS’s three full copies. It reports that this reduced file-storage overhead by 33%. The supplied material does not specify the measurement method or the conditions behind that figure.
Storage Tradeoffs at Google Scale
The examples show how storage architecture involves tradeoffs between space, speed and resilience. Storing multiple complete copies is straightforward, but erasure coding can use less storage while preserving the ability to recover data. Data structures matter too: reducing memory overhead or improving access patterns can affect the performance of systems that handle large volumes of information.
For readers building databases, the interview offers examples of how engineering choices in indexes, storage formats and data structures can shape a system’s behavior. Mattis’s account is a practitioner’s perspective rather than a formal comparison of database designs. It does not establish that one approach is best for every workload.
The AI discussion adds a current dimension to that experience. Mattis argues that tools can multiply the impact of engineers with deep knowledge of a system. The episode also raises questions about code review as more code is produced with AI assistance, though the supplied description does not report a specific change to Cockroach Labs’ review process.
Top picks for "distribut databas peter"
As an affiliate, we earn on qualifying purchases.
From GIMP to Google Storage
Mattis’s work predates his role at Cockroach Labs. He and Kimball created GIMP, the free, open-source image editor. The source says the first version of Google’s logo was made with GIMP. It also recounts that Mattis initially declined an offer from Sergey Brin because of the commute from San Francisco to Mountain View, then joined Google after another startup did not work out.
At Google, the interview links his work to Gmail’s early storage needs and to the evolution from Google File System to Colossus. Gmail launched in 2004 with 1GB of free email storage, according to the source. The episode uses these systems to discuss practical constraints in managing data, rather than presenting a general survey of distributed databases.
Mattis also recounts replacing standard library data structures in performance-sensitive situations. Beyond the C++ B-tree, the source says he developed a Swiss Table implementation for Go that was faster than Go’s map implementation and later made it into the Go library with help from the Go team. These examples underpin the episode’s recurring observation that B-trees appear in varied storage and indexing roles.
““There’s always going to be someone else working on your idea.””
— Peter Mattis, as quoted in The Pragmatic Engineer interview
Limits of the Episode’s Claims
The supplied report summarizes an interview but does not include a publication date, full transcript or detailed technical documentation for the systems discussed. The 33% reduction in storage overhead is reported without a measurement window, workload or comparison methodology, so readers cannot assess how broadly it applies from this material alone.
Mattis’s statement that AI has increased his productivity without reducing quality is a personal account. The source provides no productivity figures, independent evaluation or quality measurements. It also does not specify which tasks or time period underpin his assessment. The episode’s discussion of future code review is described as a topic, but no proposed policy or outcome is given.
The source says that a network round trip within a zone took milliseconds when Colossus was built and is about 100 microseconds today, describing that as 10 times faster. It does not provide dates or define the endpoints and measurement conditions. The supplied summary also makes a claim about routing data through space for speed; it offers no route measurements or comparison data to establish when such a path would be faster.
Where to Hear the Interview
The episode is available through YouTube, Apple and Spotify, according to The Pragmatic Engineer. The publication says a transcript appears at the top of its episode page and timestamps are listed at the bottom. Those materials may provide more detail on Mattis’s technical examples and the discussion of AI and code review.
The source does not announce a product release, database benchmark or policy change arising from the interview. For now, the next step for readers seeking specifics is to consult the full episode and transcript, especially for the definitions and conditions behind the storage and latency figures.
Key Questions
What is the news development?
The Pragmatic Engineer published an interview with Peter Mattis about distributed databases, Google storage systems and AI-assisted coding.
Who is Peter Mattis?
Mattis is a co-founder and CTO of Cockroach Labs. The source also identifies him as a co-creator of GIMP and a former Google engineer who worked on Gmail and distributed storage.
What does the episode say Colossus changed?
The report says Colossus used Reed–Solomon erasure coding, storing data twice while increasing redundancy compared with GFS’s three full copies. It reports a 33% reduction in storage overhead, but the supplied material does not give the measurement conditions.
What does Mattis say about AI coding tools?
According to the episode description, Mattis says AI has brought him back to writing code and made him more productive without a drop in quality. The report presents this as his experience; it does not include independent measurements.
Source: rss
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
