text stringlengths 0 1.49k |
|---|
**Jason Keene:** Thanks! |
**Andrew Poydence:** Thanks! |
**Carlisia Thompson:** Thanks, this was great! |
**Erik St. Martin:** Thanks everybody for listening. You can follow us on Twitter @GoTimeFM. If you wanna be on the show or have suggestions for topics or guests, file an issue on [ping](https://github.com/GoTimeFM/ping), and with that, goodbye everybody. We'll see you next week. |
**Carlisia Thompson:** Bye! |
• Pachyderm: a modern data lake built on containers |
• Version control for massive datasets |
• Data provenance: tracking changes to data and analysis |
• Applications in machine learning, particularly with EU's new regulations |
• Customer use cases include General Fusion (fusion reactor) and financial institutions |
• The system discussed is written entirely in Go and uses Docker containers. |
• Pachyderm's architecture includes a daemon (Pachd) written in Go using gRPC, and a frontend command line interface tool (Pach Control). |
• The motivation for choosing Go was due to the existing components being in Go and aligning with Google's internal use cases. |
• Pachyderm handles data orchestration, while users handle data processing within containers. |
• The system allows for complex pipelining of data sets and distribution across multiple containers. |
• Discussion about potential episode where a guest is assigned a mission to try out the system and return to discuss their experience. |
• Development of Pachyderm orchestration system in Go |
• Benefits of using Go in Pachyderm, including batteries included standard libraries and goroutines for concurrency |
• Scalability issues with large data sets (hundreds of gigabytes) and limitations of Docker containers |
• Future plans to handle larger data sizes (multiple terabytes) |
• Potential use cases for distributed file systems and discussion of existing projects like Minio and RADOS |
• Collaboration between Pachyderm and other open source projects, including support from the Minio community |
• Challenges of making money from open source software |
• Importance of aligning incentives for developers |
• Deploying open source products and navigating deployment costs |
• Case study: Pachyderm's decision to deploy on Kubernetes |
• Business models for open source projects (support contracts, hosted models) |
• Communicating vision and attracting community engagement through charismatic leadership |
• Role of project leaders in shaping the adoption curve |
• Charismatic but controversial leaders and their impact on the open source community |
• Discussion of Linus Torvalds and his role in creating a decentralized version control system (Git) |
• Critique of GitHub for being closed-source, despite its contributions to the open source community |
• Pros and cons of open source software, including the potential for centralization vs. decentralization |
• Gitea and Pachyderm as examples of open source projects challenging the status quo on GitHub |
• Vision for a decentralized data processing platform (Pachyderm) similar to GitHub's role in version control |
• Wuzz: a terminal-based HTTP request tool |
• Ozzo Validation: a Go validation package with separate rules and nested validation |
• Melissa Data: a data cleansing and validation service criticized for being outdated and using C |
• Dep: a dependency management tool in development to solve the Go dependency problem, with an article explaining its use and upcoming episode featuring Sam Boyer |
• Discussion about the math behind dependency chain and graphs in a tool called GPS (packaging solver) |
• Dependency management as a major problem in software development, with Go being no exception |
• Comparison of different programming languages' approaches to dependency management, including Rust's Cargo and Java's IDEs |
• Mention of the Gogland IDE from JetBrains as a high-quality, commercially-supported tool for Go developers |
• Discussion about the importance of good IDEs in increasing language adoption in the enterprise |
• Brief overview of other news and projects in the Go community, including Vim-go Debug and Jodosha's Delve integration |
• Francesc's video about Go 1.8 |
• #FreeSoftwareFriday segment where the hosts give shoutouts to open-source projects that make their lives easier |
+ NATS from APCERA and Derek Collison |
+ HashiCorp, specifically Vault |
+ gRPC from Google |
• Discussion of Pachyderm and its use of gRPC |
• Erik's hypothetical recommendation for password cracking with Hashcat |
• Show sponsors: Toptal and Backtrace |
• Reminder to subscribe to GoTime FM and follow on social media |
• Warning about using Pachyderm: don't flood your house with data lake |
• Episode release in a week and anticipated memes/gif responses |
• Official goodbye from hosts and guest |
**Erik St. Martin:** Alright everybody, welcome back for another episode of GoTime. It's episode \#34 today, and our sponsors today are Toptal and Backtrace. Today on the show we have myself, Erik St. Martin, Carlisia Pinto is also here... |
**Carlisia Thompson:** Hello. |
**Erik St. Martin:** And Brian Ketelsen... |
**Brian Ketelsen:** Yo! |
**Erik St. Martin:** And our special guest today is the co-founder and CEO of a project called Pachyderm, and I don't wanna give too many details because I'd like to hear him describe it in his own words. Please welcome Joe Doliner. |
**Joe Doliner:** Thanks guys, it's great to be here. |
**Erik St. Martin:** So Pachyderm... Do you wanna give everybody a brief rundown of what Pachyderm is before we get too far into the weeds? |
**Joe Doliner:** Yeah, absolutely. Pachyderm is what's called a data lake, and the other big example of a data lake that you're probably familiar with is the Hadoop ecosystem of tools. What Pachyderm is trying to do is basically build a new, more modern version of what Hadoop is in the world today. We've taken a very o... |
When you're using Hadoop, if you wanna process some data, you're probably gonna wind up writing a pig script, you might wind up writing a Java class with a map and reduce method... There's a bunch of different frontends to it. For Pachyderm there's really only one frontend for processing data, and that's a container. Y... |
We like to say that if you can put it in a container, then Pachyderm will scale it up to petabytes of data for you, because we can just orchestrate these containers and duplicate your code and orchestrate all the data into them such that it all just floats through. |
There's one other very innovative feature of Pachyderm that doesn't exist in Hadoop, and that's version control. I assume everyone listening to this show is intimately familiar with Git. Pachyderm basically does what Git does, except it does it for gigantic data sets. As your logs are coming off of your server, as your... |
**Brian Ketelsen:** So is that like full-on data provenance? |
**Joe Doliner:** Absolutely, yeah. |
**Brian Ketelsen:** Oh, wow! |
**Joe Doliner:** And provenance is one of the key features for Pachyderm. We can do very, very granular provenance. We can do it at a Perl-like file level within the system. |
**Brian Ketelsen:** That's amazing. That's truly a big deal. |
**Erik St. Martin:** \[03:43\] Yeah, this is a huge interest to Brian and I, because Brian and I spent actually two different jobs where we worked with big data, doing a lot of fraud prediction and credit scoring... A lot of that with the laws and things like that, the provenance is huge because you can cache these cou... |
**Brian Ketelsen:** Yeah, it's a big deal. I remember when Pachyderm came out - I guess it's probably been roughly two years now. We actually played with it back at the last company we worked at, and it was pretty impressive even back in the earliest days. |
**Joe Doliner:** That's really nice of you to say. I bet there were some pretty unimpressive things about it at that point, too. |
**Brian Ketelsen:** Well, it didn't do a lot. Your earliest releases were kind of tying Docker container streams together, and I was still terribly impressed with the whole idea. I'm excited to hear that things have come along so nicely for you. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.