TL;DR
Get business pricing on networking and server gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
The Pragmatic Engineer published an interview with Peter Mattis, Cockroach Labs co-founder and CTO, about distributed storage, database design and how AI has changed his coding work. The episode revisits his work at Google and argues that deep systems knowledge remains valuable when building and reviewing software.
The Pragmatic Engineer has published an interview with Peter Mattis, co-founder and chief technology officer of Cockroach Labs, about the engineering choices behind distributed storage and databases. The episode also examines how AI coding tools have affected Mattis’s work, making the discussion relevant to engineers weighing performance, reliability and code quality in large systems.
Mattis describes a career spanning open-source software, Google and the database company he co-founded. He is also an original creator of GIMP, the open-source image editor he developed with college roommate Spencer Kimball. According to the episode, the first version of the Google logo was made using GIMP. The interview connects those early experiences to later work on storage systems used by Google products.
One example is Gmail’s early storage design. The source says B-trees helped track email threads and unread counts, with incoming messages matched to threads through a search index. At Google, Mattis also worked on Colossus, the successor to Google File System. The report says Colossus replaced three full data copies with two copies plus Reed–Solomon erasure coding, reducing storage overhead by 33% while increasing redundancy.
The episode also discusses Mattis’s work on data structures. The report says he built a B-tree with semantics similar to C++’s std::map that used memory more efficiently and performed faster in the cases he examined. Later, he developed a Swiss Table implementation for Go maps that the report says was faster and eventually became part of Go’s standard library, with the Go team helping complete it.
Storage Choices at Large Scale
The examples show how seemingly small design choices can affect the cost, speed and reliability of infrastructure used by large services. Colossus’s reported reduction in storage overhead illustrates why distributed systems teams look beyond keeping multiple full copies of every file. Alternative redundancy schemes can change how much storage is needed while preserving the ability to recover data, though the source does not provide operational details or independent measurements beyond the figures it reports.
The discussion also links database engineering to practical performance work. Replacing a standard library structure can help when profiling identifies a genuine bottleneck, but the episode’s examples do not establish that custom structures are preferable in general. Mattis’s account instead highlights the role of measurement and domain expertise: engineers need to understand what the system is doing before deciding where a specialized design may help.
AI is another reason the interview may interest software teams. The source says Mattis feels more productive writing code with AI after spending time in management, and that he sees potential for AI to improve quality and extend the impact of experienced engineers. Those are Mattis’s views, as presented in the interview; the source does not offer controlled evidence that AI tools improve code quality across teams.
Top picks for "distribut databas peter"
As an affiliate, we earn on qualifying purchases.
From GIMP to Cockroach Labs
Mattis’s path began in open-source software. The report recounts that he and Kimball nearly delayed GIMP’s release after hearing about another ambitious image-editing project shortly before launch. They shipped anyway. Mattis’s advice to founders, as quoted in the source, is that others may share an idea, but many will not release a product. The anecdote offers background on his early work without establishing a broader rule about startup outcomes.
In 2001, Sergey Brin invited Mattis to interview at Google and made him an offer, according to the report. Mattis initially declined because he lived in San Francisco and did not want the commute to Mountain View. After a later opportunity, he joined Google. His work there included storage and data structures, before he co-founded Cockroach Labs, which develops distributed database software.
The interview returns repeatedly to B-trees: they appear in the report’s account of Gmail, Mattis’s alternatives to standard library structures, and CockroachDB’s range index. The source points to a paper titled The Ubiquitous B-Tree as further discussion of their recurring role. This is a thread through the conversation, rather than a claim that every distributed database or storage system uses the same design.
““There’s always going to be someone else working on your idea.””
— Peter Mattis, as quoted in The Pragmatic Engineer
Evidence Behind the AI Claims
The supplied report summarizes an interview but does not include its full transcript, publication date or independent verification of the technical figures. The reported 33% reduction in storage overhead is attributed to the account of Colossus, but the source does not set out the measurement method, workload or comparison conditions.
It is also unclear from the source how broadly Mattis’s performance improvements for C++ and Go data structures applied, or what workloads were used to compare them. His view that AI can support higher-quality engineering is a personal assessment in the interview; the report gives no benchmark, sample size or team-level results to establish how general that effect is.
The Interview and Its Follow-Up
The episode is available through YouTube, Apple and Spotify, according to The Pragmatic Engineer, which also points to a transcript and episode timestamps. Readers seeking more detail on Mattis’s remarks can consult those materials. The source does not identify a further announcement, product launch or scheduled milestone arising from the interview.
The report also directs readers to turbopuffer’s documentation of a planned storage architecture redesign, described as an effort to make search faster, cheaper and more reliable at scale. That work is separate from the Mattis interview; the supplied material does not establish a direct connection between the projects or give a completion date.
Key Questions
Who is Peter Mattis?
Peter Mattis is the co-founder and CTO of Cockroach Labs. The source also identifies him as an original creator of GIMP and a former Google engineer who worked on storage systems.
What does the interview cover?
The episode discusses distributed databases, storage engineering, data structures and AI-assisted coding, alongside Mattis’s path from open-source development to Google and Cockroach Labs.
What did Colossus change?
According to the source, Colossus used Reed–Solomon erasure coding with two stored copies instead of Google File System’s three full copies. The report says this reduced storage overhead by 33% while increasing redundancy.
Does the report prove AI improves code quality?
No. It reports Mattis’s view that AI has made him more productive and can improve quality, but provides no independent study or benchmark establishing that result across engineers or organizations.
Source: rss
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
