DSS 2026 — How to Kill Your Own Chain: A Security Playbook for Live Cosmos→EVM Migrations Ofir Eliasi Source video: https://www.youtube.com/watch?v=C6zW_4asU74 Automatic captions — not human-reviewed. Complete spoken-audio coverage is unverified. Source language tag: en (English). Source export SHA256: 598694e23a435766dd4ebd0a73a03dfbb110ab5ed1986ad439138b70883c27f9 [0:00] Um, so, hello everybody. [0:03] Um, [0:04] I'm going to talk about like a topic [0:06] that is um for me very interesting. We [0:09] uh I got involved into uh [0:12] some scenario that is um [0:15] um super complex, but I see it happens [0:17] more and more recently and expecting to [0:21] have it more on the future. So I think [0:23] it's like super interesting and [0:24] something that was having [0:26] specifically from the security aspects [0:28] um [0:29] even more. [0:30] Um, [0:32] first like short introduction about [0:33] myself and my background like over 20 [0:36] years of software development, two times [0:39] founder, [0:40] um been involved into blockchain [0:42] full-time from 2017, [0:44] many projects, most of them in [0:47] related to infra, MEV, RPC, security [0:52] um and so on, research that relates to [0:55] blockchain as well. [0:56] Um, and that's also what brings me to [0:59] this [1:00] specific case. Now the CTO of Lava [1:03] Network, an RPC provider that is [1:06] trustless than decentralized, uh [1:09] currently run on running on Cosmos. [1:12] And um [1:13] we got into the conclusion that um [1:17] this chain or this network actually need [1:19] to move from uh Cosmos based network [1:23] into an EVM [1:25] protocol, smart contract of course. [1:28] Um, so [1:31] first like very shortly about the the [1:33] background or the the reason of moving. [1:36] As you know, Cosmos based and generally [1:37] up chains involve with like many aspects [1:40] super complex [1:41] and sometimes even expensive. The idea [1:44] of of shifting is to cut the the [1:47] security layer [1:48] to get many of the elements that you can [1:51] handle without them [1:52] uh out of your scope and in that case [1:55] save a lot of [1:57] time and effort and resources. [1:59] So and obviously we're going to talk [2:01] about that uh [2:02] like broadly. [2:04] Uh I have another talk in two days, [2:07] three days actually, um [2:09] in Devcon in the main event about why uh [2:12] an upchain is not necessary or when [2:15] upchain is necessary. In our case it's [2:17] not. [2:18] Uh but for now we're definitely going to [2:19] focus on the security aspect of moving a [2:22] chain or a network [2:24] uh [2:25] into an EVM smart contract protocol [2:28] base. [2:29] So first [2:30] it's huge event. It's a [2:33] it's a [2:34] it's an earthquake to a to a protocol [2:37] that involved with many many element. [2:39] Um when you move something out out of an [2:42] upchain, it means you take everything [2:46] with you. And if it's like different [2:47] ecosystem like Cosmos and and EVM, [2:50] that means [2:52] any of those get a different [2:54] interpretation. [2:55] You like it's your token and if you have [2:57] any mechanism of staking and vesting and [3:01] roles, identities [3:03] and so on, they all move with you and [3:07] should move on the same time. [3:09] So and that's involved as I said like [3:11] with um a couple of aspects of how you [3:14] do deal with that because it's unrelated [3:17] uh ecosystems. They have no connection [3:19] with them between them. Um of course a [3:22] bridge is definitely not not considered [3:25] as connected. It can only impact for for [3:28] like not not positively on on such case. [3:32] But it's also means that you need [3:35] other people collaboration. You have a [3:36] community, you have token holders, you [3:38] have stakers, you have other [3:40] participants in the network and you need [3:42] them also to be involved with this [3:44] migration. [3:46] And they [3:47] not obey your rules. They sometimes take [3:49] their time and sometimes they not exist [3:51] or gone. Um [3:52] and that's also something that you need [3:54] to take into consideration. [3:56] Besides that other elements very [3:59] very intuitive like identity, a wallet [4:02] on Cosmos base uh network is completely [4:05] different than EVM. [4:06] And that's a challenge that you need to [4:08] handle. [4:09] So, overall all of this means that you [4:11] need to get [4:13] everything considered in advance. [4:15] Get your um steps planned and make sure [4:19] that [4:20] the elements that relate to you you're [4:22] handling them. What whatever is not or [4:24] depends on other, you know how to handle [4:27] it properly and to take into [4:28] consideration any scenario because even [4:32] when what is uh considered is like [4:34] unreasonable, like people giving up [4:36] their funds, [4:37] it's obviously available on such case. [4:42] Um so let's talk about like the exact [4:44] plan. This is again this is my [4:46] interpretation. [4:47] Um I split it into all the steps that [4:50] can be involved, but definitely you can [4:53] merge some step or do them together. You [4:56] can um [4:57] give up some of them or have something [5:00] that be different, but [5:01] the common sense or what mostly [5:04] available and others also did having [5:07] such a scenario is is considered like [5:11] it's very similar to uh what I'm going [5:12] to show right now. So first, as I said, [5:15] identity is one of the the biggest [5:17] challenges. You need to map where if you [5:19] are if you want to say that like in a [5:20] very uh [5:22] um naive way, you need to map any wallet [5:25] from the app chain Cosmos in our case [5:28] into EVM. How to do that? You can go [5:30] very straightforward and sign something [5:33] on Cosmos that will be um um executed on [5:36] chain on the EVM chain that you're [5:38] heading. [5:39] Um but that means you need to [5:40] collaborate you need others [5:42] collaboration and you cannot set them a [5:45] timeline that will be very strict. That [5:47] 's why on that case you take some uh [5:49] some gap and having this like couple of [5:52] days, sometimes week, and sometimes even [5:53] a month. [5:55] After that we're going to handle each [5:56] security aspect of those like later, but [5:58] just like briefly to go through that. [6:00] After that, of course, um shifting [6:04] everything you have [6:05] from upchain into EVM. It means taking a [6:09] snapshot involved with all the data that [6:11] is related the current up-to-date data [6:14] and and put it on chain [6:17] on the target chain. It's including of [6:20] course the token and the token holders, [6:23] stakers, [6:25] roles, identities as we said everything [6:27] just shift [6:29] into that chain. [6:31] After that once you did that and [6:34] deployed on the other chain you need to [6:36] actually deploying is actually the [6:38] loading. [6:39] You need to open the the new chain the [6:40] EVM chain up for running and go back to [6:44] business [6:45] on the EVM chain and move on from there. [6:48] One step after is to stop the the old [6:51] chain or the EVM upchain the sorry the [6:54] upchain. By the way ignoring it is also [6:56] an option but you know we cannot ask [6:58] validators to stop or you can ask but [7:00] they not need to not necessary they're [7:02] going to do what you are saying. [7:05] Just ignoring all I cancel it completely [7:08] and the last step is [7:11] the rest of the time we call it forever [7:13] when on that [7:16] time window things can also happen and [7:19] messed up. [7:21] So let's go one by one. Registration as [7:23] we said not okay not our [7:26] control we need collaboration of the [7:28] community get them to sign up. [7:30] It's something that gonna run for a [7:32] while. [7:33] Okay [7:35] that means you have some kind of a [7:37] contract or a way to map [7:40] old identity into new identity and [7:44] obviously on that area it's a huge risk [7:46] that relates to the community. Other [7:49] people [7:50] not you that gonna do the mapping you [7:53] ask them to collaborate and everything [7:56] that can be like [7:57] misinformation, disinformation, all kind [8:00] of confusion is is the is the most [8:04] significant significant risk about it [8:06] but as well is stealing identity and [8:09] thing that can be related to that. [8:12] Of course, you should limit that into a [8:14] specific time window and make sure [8:17] to get that collaboration. [8:19] We need to remember that on that case [8:22] there is the scenario that you won't get [8:23] the collaboration that you're asking [8:25] for. What you do in that case? How you [8:26] handle that? This is also part of the [8:28] scenarios that you need to handle and [8:30] have a backdoor [8:32] to somehow keep running on the the the [8:36] old chain. [8:38] But besides this part, [8:41] the rest of the parts mostly have no [8:43] rollback or no way to go back and that's [8:46] also something we need to put on focus. [8:49] The snapshot, as I said, taking like a [8:52] decent amount of data sometimes can be [8:54] like huge, so in some cases you need to [8:57] split them [8:59] or to execute that like in steps in a [9:02] way that you actually frozen the time [9:05] like freezing the time [9:07] and take it to the new chain. [9:12] Of course, like this is a step that you [9:14] actually control. You can do immediately [9:16] or on the time that you decide to do [9:18] that, [9:19] but the you need to remember that the [9:22] window between between the the snapshot [9:25] and the next one that is actually [9:27] loading that snapshot into the new chain [9:30] is the the time where most of the mess [9:33] happening. [9:34] You actually have two chains running. [9:37] You have [9:41] participants or like community members [9:43] that operating both [9:45] or none [9:46] and a lot can happen. You also have a [9:49] time where people can [9:52] run the the old chain since you don't [9:55] want to have like too much of a down [9:56] downtime [9:58] and still expecting getting rewards on [10:01] the new chain. It can be a transfer, it [10:02] can be anything like that. [10:04] We need to also remember that this on [10:06] this time frame we also cutting anything [10:09] that can be [10:10] um a confusion or a combination of both [10:12] like bridges. If we have a bridge from [10:14] the [10:15] um the old upchain to uh the new EVM [10:17] chain that need to be on hold and [10:21] actually to be uh [10:23] killed in a way that you not get like [10:25] double spend. [10:28] Um one more element that will let the [10:30] security no one actually promise you [10:32] that [10:34] uh deploying this transaction or [10:36] floating the the data will actually [10:38] succeed. Um it can be like technical [10:41] issues and bugs but can also be someone [10:44] that is like specifically trying to [10:46] target this transaction and do [10:48] everything he can in order to fail it. [10:50] Why? [10:51] Many reasons but definitely it's doable [10:55] not that hard to do that and sometimes [10:57] someone to will try to four trying this [10:58] transaction and make things in a way [11:00] that um can benefit out of it. So this [11:04] time frame is mostly uh sensitive in a [11:07] way that you need to cover the gap [11:09] sometimes afterwards. [11:13] Then last step when you open it back to [11:16] uh business on the new EVM chain uh we [11:19] need to remember one thing. [11:21] Um the idea or the motivation is to have [11:24] as minimal as downtime [11:27] as we can. This protocol is running [11:29] someone is is using it [11:31] and [11:32] in this case [11:34] um [11:35] like the idea is to decrease the time [11:37] between the freeze and opening the new [11:41] protocol. [11:42] Um again some [11:45] will will some protocol decide to still [11:47] operate on the old chain until the open [11:50] and not stop operating on the freeze and [11:54] on that case you need uh again you need [11:55] some way to make up on that time to [11:58] who's um um decide to still uh execute [12:03] and you need to incentive them. It can [12:04] be like by the foundation, it can be [12:06] manually but it can be part of a format [12:08] that later do catch up and get a second [12:12] wave of of [12:14] of deployment of data that relates to [12:16] the old chain. [12:19] Here mostly [12:21] they dare to get back to normal, [12:23] to get the users involved and to have [12:26] them [12:28] into like the routine. Remember, if you [12:30] have staking, for example, and that's [12:33] the interest of the [12:35] the protocol to keep those staking, [12:38] you definitely don't want to get them [12:39] out of staking and ask them to restake [12:41] because [12:42] it's going to be involved with the [12:43] churn. People not necessarily all of [12:45] them will stake again. So the best idea [12:49] on this case is to keep all to copy [12:53] every element or every aspect of it [12:55] into [12:57] the new chain. By the way, in in Lava's [12:59] case, [13:00] what happened is that we besides [13:03] staking, we also have vesting. [13:05] And on the Cosmos based network, we [13:07] allow to who has token that are vested [13:11] to stake and to participate in the [13:14] in the protocol process. [13:16] So in that case, [13:19] it's the idea is to take everything that [13:21] relates to that and put it on chain and [13:23] keep and preserve the state of staking [13:25] and participation in the the the process [13:29] of the protocol. [13:30] Last one, [13:32] of course, stopping is not mandatory or [13:34] should be defined according to the [13:38] the involvement of [13:41] the community, but definitely ignoring [13:43] is also an option. We just need to [13:46] define it like officially that this [13:48] chain is irrelevant anymore. On that [13:50] case, [13:51] many can has funds lost, but you can [13:54] handle that technically by stopping or [13:56] having a new version that will stop any [13:59] functionality that relates to the chain. [14:01] But again, that's not not like the [14:04] biggest challenge we have. [14:06] Last one, [14:08] since then, after we open and stop the [14:10] the old chain running forever, it seems [14:13] like what the what issue can [14:15] uh what we can have on [14:17] uh like the day-to-day ongoing on the [14:19] new chain. Well, remember that not [14:22] necessary everybody claimed their [14:23] identity, not everybody claimed their [14:26] funds. And that's still [14:29] um something a scenario that we need to [14:32] um consider because [14:34] um [14:35] eventually someone can get this funds or [14:39] try to hack them and get them um [14:42] like instead of those that actually own [14:45] them. [14:46] Um [14:47] yes um so we kind of ran out of time. So [14:52] I think like the main message from this [14:55] um [14:56] talk is [14:58] definitely [14:59] um um the awareness of having such steps [15:04] and the the ability to split to split [15:07] the the process into time frames and and [15:10] time windows [15:11] and to handle each one of them [15:13] separately. [15:14] Um [15:15] um so the idea is like of course the the [15:18] to kill the chain but not kill the [15:20] users, make them stay and keep the the [15:22] protocol running on the new chain. [15:26] Questions? [15:29] Good, we ran out of time. [15:30] Um so about that my my schedule is like [15:34] definitely I have one more this today [15:37] later, but as I said I'm going to talk [15:39] about this topic like more about the [15:41] reasons and how to handle it technically [15:43] later this week. [15:45] Thank you everybody.