Interview with Dahlia Malkhi, founding member of VMware Research
When people hear the term blockchain, they think of Bitcoin. However, that would be similar to thinking that the internet is the same as email. Cryptocurrency is a single use case for an underlying technology that has many use cases. The initial use case for cryptocurrency has had dramatic impact in finance around the world, allowing the transfer of value from one person to another through a decentralized means, bypassing age-old behemoth institutions.
During 2017, the cryptocurrency market saw unprecedented increase in use and general acceptance. In March 2017, the value of one ounce of gold was on par with one Bitcoin. In December 2017, one Bitcoin was equivalent to 16 ounces of gold.
There are a number of factors that caused this, such as speculation and gold-rush mentality. However, there is also inherent value with some problem being solved. That is the reason why there are so many different currencies and tokens out there, each solves a different problem.
The technology that enables Bitcoin is called blockchain. It allows for a distributed ledger (or database in other words) that maintains consensus while getting data from untrusted sources. The use cases for this are so vast that you can think of now as the dawn of a new age of computing that will usher in change more dramatic than the introduction of the internet in the 1990s. In the coming years, you will see a meteoric rise in new blockchain use cases similar to what has occurred with Bitcoin.
In January, at the VMware HQ in Palo Alto, I had the opportunity to interview Dahlia Malkhi, the principal researcher and founding member of VMware research. Dahlia’s department is heavily invested in researching new and novel methods for advancing the use cases of blockchain. Here is the transcript of that interview.
Video podcast:
Audio podcast:
Podcast transcript:
Daemon Behr: [00:01:53] Thank you for taking the time to sit down with me for this podcast. I have a number of questions, but first off, before we get into the conversation I was wondering if you can tell me a bit about your background and what is it that you do at VMware.
Dahlia Malkhi: [00:02:07] Sure. I’m Dahlia Malkhi, I’m a researcher, part of a research group at VMware at the office of the CTO. We were established three years ago in 2014 when VMware felt that in order to meet threats of a future where a lot of workloads migrate to the cloud, it (VMware) needed to put a lot of emphasis on innovation. It’s doing it in a variety of ways, and research, a dedicated research arm is one of the ways in which VMware drives innovation.
[00:02:43] There is obviously innovation and inventiveness in VMwares culture throughout the entirety of the organization right from the outset. And I personally I’m a researcher with background in reliability, security, foundations of distributed systems and with activity and contribution for longer than I want to admit, but let’s say above two decades. My interest is all algorithmic areas of distributed systems, reliability and security
Daemon Behr: [00:03:17] Alright, fabulous. Thank you very much. So today I’d like to talk about blockchain technologies and distributed ledgers. And I was wondering if you can just give me a background on some of the technologies and how they all fit together.
Dahlia Malkhi: [00:03:31] Yes, so, let me give you the two-minute definition of the technology and then branch into a number of directions like use cases and what are the technical underpinnings of the technology. So, in a nutshell what is a blockchain? A blockchain, is a technology that allows collaborative decisions and information sharing among participants that don’t necessarily trust each-other. And it allows them to execute collaborative logic, business logic without actually trusting each other. So, it allows you to construct services that previously were built on your ability to trust. An institution, or corporation. The business incentive that they have in supporting such a service and their own reputation and their own experience. This was all behind the trust you would put in a service. What it allows you to do, is the same thing without putting a trust in any single institution or government or service behind it. So that’s, that’s, the two-minute definition I can give you.
Daemon Behr: [00:04:42]Great.
Dahlia Malkhi: [00:04:43] And now we can dive into any technical direction you want.
Daemon Behr: [00:04:46] Sure. So blockchain has become top of mind to a lot of people and it’s more so with the public than ever with the rise of crypto currencies. Can you tell me a bit about how we got to where we are right now and what are some of the use cases?
Dahlia Malkhi: [00:05:04]So blockchain emerged as the technology that allowed a cryptocurrency platform to track the provenance and the activity of digital assets and it was developed and launched to solve that particular problem. How do you track the activities of digital assets, prevent the double spending of you know any digital value and reach agreement with public participants out there on who owns and where the digital asset is and who has permission to spend it and use it. So, this is how this started this was of course populated and publicized by bitcoin. It is still a very, very good technology for doing just that for launching a digital coin, coupons, whatever you want to call it and jointly tracking the provenance of those coupons. Everybody knows where they go, where they are, can track their activity and how they are being transferred. And this led to a whole industry around just generating coupons and getting value out of it. But if we double click if we zoom on the technology itself and the basis of this technology, is the idea of forming agreement on the history of actions that participants in a system would like to introduce to any kind of digital form. And in fact, this is a problem or a challenge that is not new in the technical world academic world and industrial world has looked at technologies that allow distributed services to be constructed such as this agreement. And there’s agreement, and there’s replication there’s redundancy in the copies of you know the history of the updates that you introduce to a particular service. And also, a variance of it where the participants don’t even trust each other or assume that some of them might be compromised and that some really penetrated by a hacker and be controlled by malware. So those technologies are referred to in the technical literature and in the technical community as Byzantine Fault Tolerant Technologies or BFT. And again, at the basis of them, they solve a very well-defined and very concrete problem. How to reach agreement. How to form agreement. In a distributed system among multiple participants on history of updates to a particular domain.
Daemon Behr: [00:08:22]What are some of the mechanisms that are used to defend or prevent the double spend issue? How does that actually work?
Dahlia Malkhi: [00:08:37] Once you have a public global agreement on a ledger or a sequence of actions which are done and performed over your account or your digital assets, that’s it. The problem is solved. So really at the core of all this technology is the ability for us to share this information and not to argue about what happened. So really, the double spend problem is that you have a digital coupon of some sort. It’s very easy for you to make a thousand copies of that coupon. You try to spend it. Over and over and over. But once we have one place and just like normal banks or clearing houses have today in the physical world when you have a place where the only way you can spend that coupon is by recording it on the ledger which is global and shared and agreed on everybody. That’s it. Problem solved. And again, this is why the core of a blockchain is forming agreement and sharing this ledger. This history of activity over and I’m intentionally leaving it opaque. What is this asset? What is this information? What are these updates updating? The original one, the bitcoin blockchain ledger recorded the transaction in bitcoins. But more generally this could be anything that you can express in digital form.
Daemon Behr: [00:10:14]How is the consistency of the blockchain managed? How do, for instance with bitcoin, miners come into play?
Dahlia Malkhi: [00:10:22]Yes now were getting into the solutions and there are really a number of categories the solution. The Bitcoin solution is a permission-less solution that allows an unspecified population of unspecified size of participants to all form this agreement. The way it works is using a crypto technique known as crypto puzzles or proof of work. It’s using computation to moderate the pace in which the proposed activity to the global Ledger is throttled and is managed and then everybody is incentivized to continue the longest standing ledger that they know of in order to win transactions in this ledger. So, this is a permission-less solution. It’s very scalable and it’s very democratic. Nobody needs permission to enter this. Some of the downsides of this solution is it requires a lot of energy to participate and it led to this arms race, where in order to win this proof of work challenge you need to spend a lot of computation and there’s a strong economic incentive. So that led to a lot of energy being dedicated to solving these puzzles. It’s also, at the end of the day not a solution that has any guarantee with it. Not just moving forward but also looking backwards and there’s no guarantee that at any time if the computation that you spend at a particular moment dominates the system there’s no guarantee that you cannot revert back any activity that was already considered accepted or finalized in the past. So, this is a technology that is very interesting, very scalable, very robust so far, but it has these downsides. So, this is one category of solutions. People are very, very excited about the functionality that the technology opened to raise awareness to. But we’re looking for alternative solutions that don’t necessarily use proof-of-work. That have better finality guarantees and that you know maybe have better control over. And there’s a spectrum of such solutions. There are solutions that borrow from the standard BFT literature where the group of participants is actually permissioned. It’s very well sanctioned. You can think of a consortium of organizations that join forces together and decide in a group to solve this problem. So, you know who participates, you know their number. They don’t just come and go at will and the type of solutions, the type of BFT mechanisms that are possible in that setting are very, very different from the Bitcoin blockchain.
Daemon Behr: [00:13:47]So since they are more structured and controlled, does that mean that there is less energy required to provide the proof of work or do they use another mechanism in order to continue the chain?
Dahlia Malkhi: [00:14:04] Yeah absolutely. So, they don’t rely on energy at all. Work has traditionally been the safety keeper for such settings. It’s the assumption that some fraction of them is in fact honest and not corrupt and not penetrated. So, this is one way in which they can guarantee safety as opposed to proof of work. And there’s actually a spectrum of other solutions. So, to marry together the economic incentives with the guarantees of BFTs, people have started looking at Proof of Stake. Where it is a permission setting you’re working with mechanisms that are essentially designed for a closed permission group but the permission is not really just stop by assumption or by design. It’s something that has economic power behind it. So, the way you get permission to join the group and also to behave correctly in the group is tied directly to a stake that you put in participation in the form of actually putting a deposit of the relevant cryptocurrency and mechanisms to enforce that you lose this deposit, if you are detected with any kind of misbehavior. Typically, the idea is that the stake can be much, much higher than anything you stand to gain before you know being detected with very high probability. So, there’s as economical proof, if you like, that you have no incentive to misbehave you lose more than you gain.
Daemon Behr: [00:15:55]Are they a number of organizations or enterprise consortiums that are currently employing this method right now?
Dahlia Malkhi: [00:16:04]There are. I wouldn’t say organizations, but there are quite a few cryptocurrency startups out there that are already employing proof of stake mechanisms at Tendermint is one that comes to mind. Ethereum has announced recently that in 2019 they’ll be moving gradually into a proof of stake mechanism that is actually operating in that very interesting hybrid mode where the main ledger remains the public, permission-less ledger. And then there’s the proof of stake engine which finalizes and provides an extra stamp of approval for that public chain. So, it becomes the authority whereas the public permission-less Ethereum chain is essentially going to become a proposal mechanism. And down the line, they’re also talking about completely switching over to proof of stake mechanism.
Daemon Behr: [00:17:08]Since a lot of coins in the cryptocurrency market are based off Ethereum how will this change in strategy, or essentially “a fork of in Ethereum”, effect these other coins?
Dahlia Malkhi: [00:17:22] So this is a very interesting question. So far there hasn’t been a colossal collapse of Ethereum or bitcoin or any of the other cryptocurrencies. So, in some sense, no news, it’s just going to continue as usual but maybe provide some protection down the line from something that might have happened. I believe that the full ramifications on the ecosystem and who will be stakeholders and what exactly, how exactly it will play in terms of incentives and rewards. Those are probably not fully understood yet and time will tell.
Daemon Behr: [00:18:05] It seems like there’s a lot of emerging use cases for blockchain technology now that it is becoming more and more common. Things such as supply chain tracking and so forth. What do you see as future use cases that would be beneficial or perhaps are emerging right now?
Dahlia Malkhi: [00:18:28]So this is terrific, and probably a billion or trillion-dollar question that the entire venture capitalist community would love to get an answer for. A lot, a lot of money a lot of hopes and activity is put into this space. I can give us a somewhat technical answer. This is perhaps not the business answer that you were looking for. It definitely won’t tell any anybody who listens how to go make money from it. But from a technical point of view, I view the technology as providing an opportunity and a mechanism that allows you to remove trust from a service that you construct in four levels. And really this is the process that everybody I think has seeing the excitement about this. So, at the bottom-most level we already discussed the ability to share a ledger of transactions asset transfers of a very well defined and prefixed format for cryptographic money, coupons. And that’s very, very nice. And like you said there’s a whole industry of ICOs now, living off of that. But that’s really just the bottom-most level of what the technology is capable of. Once you are aware of this ability to form a service where you share not just the final result of a money transfer or currency transfer that you decided on. You realize that first of all you can agree on a sequence of transactions that are much, much more complicated than just “A” transfers of some amount to “B”, which is really what the bitcoin blockchain allows you to do.
So, the next level is a level where the service that builds collaborative logic shares slightly more into advanced forms of transactions than just you know “A” transfers some amount to “B”.
[00:20:52] And to give you a completely silly example but from a computer science point of view very powerful one. One I can share is something like a queue. OK we can regulate arrivals so that if you and I arrive at the same time and append something to a blockchain then you will get number 14 and I will get number 15. So, this is actual action that we put on the chain instead of being the final form of: okay you’ve decided to pay somebody else and this is what you pay. You actually let the chain decide for you what is the next ticket number and if two of us arrived at the same time you will get 14 and I will get 15. But you’ll notice that there’s already something active going on the blockchain itself. All I said is I want the next ticket. You didn’t know what it is you didn’t know what amount what number is. You essentially not put an active action on the chain itself, on the blockchain and it gave it back the result. In this case was a very silly, simple example. The action was just an increment. We have a single counter you have incremented in it. So, it was 13 before you arrived and now it is 14. When I arrive, I get the current value which is 14 and increment to 15. But from a computer science point of view what that means is that I can put any type of action and the blockchain decides on the ordering and which we execute these pre-specified actions and these actions to be as expressive as you like. So, this is level 2. We were at level 1 before, this was level 2 and we don’t stop there. We get to smart contracts.
[00:22:31] So what are smart contracts? Just like we can pre-specify very, very simple transactions, we can pre-specify slightly more elaborate transactions.
The next level is what we pre-specified is a full Turing machine execution engine which every block in block executes. And you know the user decides what’s programmed to actually execute. So, it’s you know it’s completely flexible completely programmable. Now what you get to is that the blockchain orders the sequence of actions which are executed. But what these actions are is completely flexible and completely programmable. And this allows users to build shared execution engine in a completely programmable and completely flexible manner. Without any single service behind them, without any provider, any authority or anybody being able to control or prevent or revoke or reject or you know, censor anybody else. And this is the ability to execute smart contract logic and the blockchain orders and forms agreement on what is executed and what are the results.
So, this is layer 3, and remember I told you that were four layers. So, we’re gradually removing trust from what’s traditionally services needed to execute in a centralized manner before. So, we’re now able to build this complete service from scratch without trusting anybody in a completely collaborative manner. But one thing is still missing and that is security and privacy. And this is something that traditionally you need to trust a mediator or a trusted third party for bootstrapping your key management for managing privacy and so forth and so on. And this is the fourth layer perhaps, and the most exciting is that the blockchain can actually allow you to manipulate and process private data and to prove the results of operations and manage credentials without bootstrapping by any centralized authority. And we see a lot of activity in that layer as well.
[00:24:48] So again four levels. Traditionally the only way to build software services was for a corporation, to put its reputation its business incentive and its strength in supporting it in order to bring up those services. And at all four levels, you can now defer all of this implementation to a blockchain. This is from a technical point of view what are the actual use cases whether it be supply chain, health organizations, sharing data for joint analytics. I don’t know, IoT devices, management, identity management. You know there are 4000 of them and more popping up every day.
Daemon Behr: [00:25:38] So VMware is actively doing some research into this, as it is quite apparent. Is VMware contributing back to the community via the work that they’re doing on any open-source or collaborative projects? Or is there just some internal research and vision going on within the organization?
Dahlia Malkhi: [00:26:02]VMware is very well positioned in this space for three reasons. One is, we have a lot of relevant customers in the enterprise world, the financial world, governments. We drive the IoT and workloads of the world. Secondly, we (VMware) have experience and credibility building distributed software and deploying it and selling it. We’re not only operating services from our own cloud and own services. We actually built distributed systems. And thirdly we have the research team of a team of experts with combined expertise credentials contributions in the relevant arenas for over a combined hundred years research. And these systems are not for the faint of heart. You really need a lot of technical strengths in order to get them right. So, VMware is wonderfully positioned to lead in this space. What we hope to be is the most secure, the most reliable, the most expressive place to run any blockchain workloads that there will be out there. We do publish a lot of our work and we will also have a lot of offerings in the open-source space as well as our own tools and proprietary technologies.
Daemon Behr: [00:27:36]I believe that there is a blockchain related VMware fling that’s out there right now. What can you tell me about that?
Dahlia Malkhi: [00:27:43]That fling that mostly has to do with management of you know, said distributed tools. One of the strengths and experiences that VMware can offer in this space, is the experience and strength in managing distributed software. Not just our own but also third party distributed software. So even just deploying, managing, to doing dev-ops, orchestration and monitoring of distributed software is a very, very tough problem. And this is what that fling is doing. Its doing it for the Linux Foundation hyper ledger technology and it’s contributed as an open source tool.
Daemon Behr: [00:28:25] Fabulous. Well thank you very much for your time. It’s been wonderful speaking with you. We definitely look forward to seeing what VMware is going to be announcing hopefully sometime in the near future. Thank you.
Dahlia Malkhi: [00:28:38]Great thank you.
Designing Risk in IT Infrastructure 