The “Grid” is Falling…..The “Grid” is falling
Today is a day that web hosting provider Media Temple would most likely want to forget. The web hosting company launched their new “grid” server offer which promised to be “The last hosting plan you’ll ever need.” Media Temple’s offering got a huge boost by Web 2.0 site TechCrunch in this post. The ironic thing is while TechCrunch was praising the “grid” and being hosted by Media Temple they were actually on their own servers and not the grid. The TechCrunch web site stayed up during the entiire time their readers sites were down. How much downtime did “grid” server client suffer, enough that Netcraft noticed. We’ll take our own monitoring service’s stats though which said the grid has suffered 17 outages in November with 3 hours and 10 minutes downtime, you can read the ticket by Media Temple here.
Now there have been some major discussions on Media Temple’s “grid” offering a here here here and a thread about LayeredTech’s thegridlayer.com here which is powered by vendor 3tera. One thing i especially wanted to note was this thread here in which you see me post a statement to Vlad Miloushev CEO of 3tera noting that he blew through $16+ million of VC funding when he was CEO of Z-Force which crashed and eventually became attune systems (with more VC funding [they secured 14 million November 14, 2006]).
In my recent interview with Tim Dorr of A Small Orange I asked him about Media Temple’s “grid” and he responded by saying that it was not a grid but a cluster of servers and I have to agree with him on this (as would most people). The entire thought behind a grid was that you can use the computing power behind all the nodes to solve a single problem and if any one node failed there would be no impact. We’ll let IBM define a grid:
Grid computing enables the virtualization of distributed computing and data resources such as processing, network bandwidth and storage capacity to create a single system image, granting users and applications seamless access to vast IT capabilities. Just as an Internet user views a unified instance of content via the Web, a grid user essentially sees a single, large virtual computer.
You can also view more definitions of what a grid is here here here and here with example of grids. Once you read the definition of a grid and then look at Media Temple’s grid or 3tera’s Applogic you’ll notice they are not grids. They cannot distribute processing tasks across multiple nodes and if the node you are on crashes you will go down.
Why does all this matter? Because it appears that 3tera is knowingly making false claims about their product using buzz words while subtley dropping hints that they will be looking for VC funding and Media Temple is using the buzz words to make their offer more attractive to customers in the hosting market in order to grab market share. One thing is for sure in this entire mess, the winners are the marketing departments for both companies with the losers being their customers.
November 29th, 2006 at 11:10 pm
Very interesting post.
November 29th, 2006 at 11:39 pm
Thanks Trey,
Digg it if you like the article.
http://www.digg.com/software/The_Grid_is_Falling_The_Grid_is_Falling
November 29th, 2006 at 11:40 pm
Hello,
Could you clarify why you feel that 3tera is “making false claims”? More importantly, are you aware that websites on AppLogic aren’t tied to any particular hardware node? If the node on which you’re currently residing goes down, you site will automatically be relocated to other available resources.
You seem to imply that MediaTemple’s recent outages disproves AppLogic’s value proposition, but how? MediaTemple’s platform and AppLogic are two completely separate solutions, so is it fair to draw conclusions about 3tera customers’ experience based on MediaTemple’s problems?
November 30th, 2006 at 12:22 am
Hello Isabel,
Glad you decided to comment and congradulations on your recent joining of 3tera’s advisory board:
http://www.thehostingnews.com/article2832.html
Sure I would love to clarify why I believe 3tera is making false claims. The simple answer is they are marketing a software product as a grid solution when it isn’t. Specifically the nodes can’t work together to solve a single request, in tradition it is not a grid.
In the LayeredTech implementation it is not even highly available as they have been marketing it, I believe Todd said there is a three-five minute time between when a hardware node fails and when the application piece gets moved to another node.
The Applogic product also has multiple single points of failure which is exactly what grid computing is suppose to prevent as well.
I would have mentioned Mosso in this article but i don’t know enough about their infastructure to comment but I’m guessing it is a cluster that they are calling a grid as well.
Grid is the new buzz term in the hosting industry and we are going to see many more “grid” products being launched pretty soon.
November 30th, 2006 at 3:10 am
[...] Original post by Ross [...]
November 30th, 2006 at 9:17 am
dHi Ross,
Rather than guessing, I did a search on Mosso’s website and found that the company does not call its service a grid. On the other hand, I did read on Netcraft that Mosso has had a bit of recent downtime.
You seem to feel that no company that calls its platform a grid can possibly be legit. The use of the word “grid” alone guarantees that all providers’ customers will inevitably have bad experiences, no matter how similar or different their technologies may be. But Mosso doesn’t seem to conform to this logic: they’re not a grid, and yet their customers suffered during outages?
In addition, you say you’re not familiar enough with Mosso to comment, but have you had any hands-on experience with AppLogic? I would encourage you to sign up with The Grid Layer or UtilityServe and try running some sites on our platform. I look forward to hearing your feedback on how you like *hosting on* and not just hearing about AppLogic
November 30th, 2006 at 10:26 am
Isabel,
You are absolutely right that Mosso label’s itself as a cluster and not as a grid however many sites have labeled it as a grid, something I wanted to touch on but like I said I don’t have enough info about their infastructure. You can note netcraft called it a grid here:
http://news.netcraft.com/archives/2006/10/18/extended_downtime_for_mosso.html
Lets get one thing straight, I feel that the company you are working for is making false claims by calling it’s product a grid solution. Please go ahead and read the links I provided that defines what grids are. “Grid” is the new buzz word in hosting and it is currently being abused by companies like yours for profit.
No I don’t have hands on expirience with Applogic but I don’t have to, to know it is not a grid. Even by your own CEO’s statements it shows that it’s not a grid. I do believe it is a very interesting product and can probably help hosting companies more easy manage resources/customers and limit downtime during hardware failures but that doesn’t make it a grid.
November 30th, 2006 at 10:46 am
Isabel,
Since you are commenting on this article could you please comment on why as a reporter you think it is ethical to join the advisory board of a company that you have done some extensive mentioning on in your blogs. Is it just a coincidence that a company you extensively mention in your articles later retains your services or did they pay you to mention them before you announced you joined their advisory board? I also thought it funny that you gave softlayer a hard time for using buzz words when the company you are now “advising” is doing the same thing:
http://www.thewhir.com/blogs/isabel-wang/index.cfm/2006/11/24/Lets-All-Pick-on-SoftLayer
November 30th, 2006 at 11:34 am
Ross - I am NOT a reporter. As I mentioned before, my role at TheWHIR is no different from your role at Web Hosting Talk. I am not affiliated with TheWHIR and receive no compensation for my blog posts, which express nothing but my completely biased opinion.
How did I come to be on 3tera’s advisory board?
1. Vlad posted a comment on Nicholas Carr’s blog. I clicked through to 3tera’s website, thought it sounded pretty cool, and mentioned it on my blog.
2. 3tera saw my post and offered to give me a demo.
3. We started informally exchanging ideas. I blogged about some of our conversations - not because I was being paid, but because they were interesting.
4. After a couple of months, we decided it might make sense to have a formal business relationship.
I’m sorry if you feel this is an unethical sequence of events. I’ve met many, many other people in the same way. If you were me, would you refrain from ever collaborating with any of these folks, simply because you’ve blogged about them?
By the way, I think you misread my SoftLayer post. My intention was to highlight what a smart guy I think Lance Crosby is. At the moment, SoftLayer offers only individual dedicated servers. I’m sure you’ll agree that dedicated servers are a little less grid-like than Mosso, for instance? That’s why I asked SoftLayer for clarification.
And speaking of Mosso, it sounds like you’re holding it against Mosso that Netcraft calls their infrastructure a grid? It is really fair to blame Mosso for Netcraft’s claims?
Last but not least, thanks for believing that AppLogic is interesting and helpful! But as a computer scientist, should you really believe what you read without experimenting with the technology yourself?? I went to MIT, and most of my college friends are much more interesting in getting their hands on stuff than reading documentation (much less blog posts and forum discussions). I, on the other hand, got a political science degree, so unlike you, all I can do is talk the talk
November 30th, 2006 at 11:43 am
Thanks, Ross.
I’m thinking about moving to Pair as I’ve heard great things about them from a number of people on the WWWAC list.
That said, I’m a sucker for good design and a strongly architected brand, so I may investigate A Small Orange.
-jf.
November 30th, 2006 at 5:04 pm
You certainly do get what you pay for. If you want great service you are going to have to pay the premium price for it. Note though, that with MediaTemple the $20 dollars a month is a rip off for the service they truly offer.
I used their grid server for 4 days before I decided to leave; I wouldn’t recommend it
November 30th, 2006 at 5:22 pm
Hmm, interesting article.. I was actually seriously thinking of moving to Media Temple… but I may reconsider that… I was willing to overlook the fact that their customer service could not answer any one of my questioned except by quoting from the site, even questions that could nto be answered from their ‘hype’ they still sent me a canned quote from the site in response to certain questions… I was willing to overlook this because the hosting was/is relatively cheap and it seemed like a great idea
November 30th, 2006 at 5:27 pm
[...] There is an interesting post over at HostGuru about Media Temple’s Grid. This is the company’s new hosting service, the same one that this blog runs on. I noticed a long outage about two or three days ago, which was really not great because I had linked a few people to my blog that day. The post points out a number of others outages, including one that lasted over three hours and it has some great links to disucussions about Media Temple and other grid solutions. [...]
November 30th, 2006 at 5:28 pm
MediaTemple has been good to me for several years. I’m not doing anything huge so I downgraded from a dedicated to a VPS to shared some time ago. I naturally went for the Grid Server when it was available, and I, too, have been rather disappointed with the downtime.
However, when it is up, it’s very fast. I’m hoping they iron out the kinks. Since I have nothing too major going on right now, I’ll stay with them for now. I will agree that this premature launch is causing a lot of problems, but I trust them to fix it. And if my trust is misplaced… well, then I’ll go elsewhere.
November 30th, 2006 at 5:29 pm
Ross;
I am in no way affiliated with any of the companies mentioned, although I am a MT Grid Server customer. I also administer one FreeBSD two Debian dedicated servers.
Isabel replied promptly, politely, and factually. Your attacks on her integrity make it appear that you’re the one with an agenda. I’d like to hear more facts and less conspiracy.
John
November 30th, 2006 at 5:29 pm
I think the grid has fallen and Media Temple is over priced anyways. Everyone should host their websites using DreamHost which happens to have the best pricing and best support in the industry. Ms. Wang does have some questionable ethics which is not uncommon but the choice is simple. Pay $20 per month for crappy service or pay $7.95 per month for 200GB of disk space and Terabytes of monthly bandwidth. Its a no brainer.
http://www.dreamhost.com
They are the hosts with the mosts!
November 30th, 2006 at 5:52 pm
I don’t think you will be hearing back from her, after reading your SL link about buzzwords….LOL
OWNED!
Dugg it!
November 30th, 2006 at 6:06 pm
Still, gris is not grid unless it is grid. buzzzz
November 30th, 2006 at 6:07 pm
True grids are used to implement utility computing, when applied to the xSP environment. Utility computing is just as it sounds, when your app requires additional resources, such as additional compute power, the grid can schedule additional cycles to your service. Think of it like electricity from your local electric company. Your business starts out the using low cycles, but as more employees come in, turn on office lights, computers, etc., the demand for more electricity rises. At the end of the day, your employees shutdown their computers, turn-off office lights and go home. The electrical utility was able to supply the juice, and you didn’t even notice the load as it rose.
It’s this way with grids. The problem is, you really can’t implement utility computing without specialized apps and truly clustered processes. And while much of the middleware out there will support clustering, it won’t support true grid. The LAMP stack the MT provides will not, by default, take advantage of utility computing or grid. It won’t really support clustering unless you hack dns or use some added code.
Then look at the grid market. Even the folks that claim they support true grid can’t really implement it correctly.
So I don’t know why geeks everywhere were bowled over by MT’s cheap grid offering and I would wonder, given the outages, if each node is running standalone.
November 30th, 2006 at 6:10 pm
Well, I just migrated into MT just in time to experience these performance issues, which actually started at the end of last week.
But, despite the reliability issues as MT rolls out their new server, I do think their approach has merit for handling the bursts in independent sites experiencing major traffic. Whether it’s a true grid or not, MT has a major edge over other services — and believe me, I know, with just 12,000 uniques per day I’ve gotten booted from two other hosts. I don’t think you can expect 100% uptime from anyone charging $20/mo. Once they’ve gotten the kinks worked out, I think MT could perform much better than their competitors.
You’re also completely leaving out the customer service angle. MT has 24/7 voice support, fast-responding email support, and they actually listen to their customers. I’ve been in touch with them through all of these issues and they’ve been very responsive. They’re also refunding significant chunks of service to customers who experienced this issue.
I’m not applauding server downtime. But no host is perfect, so to me one way to measure how good a host is is to see how they respond to downtime. (Heck, YouTube was down the other night for maintenance and Gmail regularly checks out!)
So, Ross, whatever argument you want to have with Isabel, as an independent web publisher I have to say I *agree* with her on several of these points!
Is it a true grid? Maybe not. (I have more of a problem with them referring to Grid Performance Units as GPUs, since that creates confusion with graphics processing units!) But I still think MT is worth a look when you’re considering services, particularly having been burned before.
And no, I don’t work in any way for MT; check my IP if you like.
November 30th, 2006 at 6:25 pm
Both of these companies were made and broken by Digg. Typical digg crap. I predict these 2 bonehead companies will not be around by 2009
November 30th, 2006 at 6:59 pm
Ross and Isabel…you both have made excellent contributions to this discussion. Maybe we could stop the flame mails and instead focus on technology analysis?
Thanks!
November 30th, 2006 at 7:13 pm
I don’t get all these “grid” plays.
As soon as you need 1-2 boxes, why use a grid?
Most of them are just Xen Virtualization. So what, they give you a few different “slices” of a slightly bigger box, on several different boxes? (or, switch you over to one in case of failure - I think that’s your main point about them not being a grid)
Btw - is Engine Yard a “grid” in your estimation? I’m impressed by their offering in the Rails area.
But still, I’d rather roll my own “grid” from inexpensive dedicated boxes from LayeredTech, and hire an admin to customize the heck out of the setup for my exact needs.
November 30th, 2006 at 7:14 pm
Thanks all for the replies:
Isabel:
My noting that netcraft called Mosso a grid was just showing how much the word is being thrown around, no it is not Mosso’s fault.
I don’t understand how looking at the front end by using a VPS from thegridlayer can help you better understand Applogic backend?
Anyway thank you for your contribution to the discussion.
November 30th, 2006 at 8:05 pm
What is the name of the game that Tim Dorr would call a so-called “real” grid application that was designed to do just what Media Temple were doing? Tell me if my “real” cog-sci PCA feature-vector driven repertory-grid application is evolving to do this too. My multi-instance “real” grid is based on my Winamp front-end, called Ingrid. Surely any “real” grid application that referenced US Patent 6825839, could also hold together groups of friends or armies of social network listeners in a radio-station format. Imagine in my idea of a “real” grid, everyone having a Universal sugar-cube sized music-catalog and my “real” grid was synchronizing their Winamp playlist, within a few minutes of starting.
I’ve had difficulty convincing people to register for my beta testing program. I’ve floundered around not actually finding a need to fill, or how to define in code what a “real” grid was, but from reading this thread, I now gather that a great way to provide incentive for users to register their Ingrid instances, is for them to provide their own content, satisfying their need for web hosting with my host-migration network scheme. My evolution so far has only conceptualized a Host Control Migration that is handed off to the next instance at the start of each song, exactly when the second hand reaches 60.
Coincidently this is the arbitrary cut-off point for the last song playing. After 10 seconds, when the new song has raised its volume from a lower preset by 50% the finishing number starts a 20 second drop from its upper to lower preset. The Ingrid instance that just finished playing now waits. On every 20 second mark during that wait a song is selected. For the first selection where the bpm/genre range is suitable the median tempo of the playing instance is adjusted over the next 20 seconds.
Shared music requests is the glue that I’m not betting money on, but they are what will hold this radio network of friends together. And yes, “there is a three-five minute time between when a hardware node fails and when the application piece gets moved to another node.”
November 30th, 2006 at 8:30 pm
Ross,
The catagorization of new technologies is among the hardest tasks startups have to contend with. Second only to actually building a working product. Despite your implication, it would be suicide for a startup to base its branding on buzzwords. That’s a game only big comapnies with big budgets can play, because buzzwords ebb and flow with the hype cycle. Startups simply can’t afford to rebrand themselves every three months and still build any credibility with customers.
The most often cited early definition of grid (google: what is the grid) comes from Ian Foster and can be found at http://www-fp.mcs.anl.gov/~foster/Articles/WhatIsTheGrid.pdf.
“I suggest that . . . a Grid is a system that:
“1) coordinates resources that are not subject to centralized control …
“2) … using standard, open, general-purpose protocols and interfaces
“3) … to deliver nontrivial qualities of service.
Ian’s definition was a good attempt to catagorize what grid computing was at the time; a system for high performance computing problems being worked on at universities and research labs. Using the grid required specialized skills, new APIs for all your code, and individual agreements between all the parties involved. Performance wasn’t gauranteed, nor was completion of your job quite frankly. Oh, and it was useless for transactional applications like those typically behind most web services.
AppLogic certainly meets Ian’s original grid definition. Our grids don’t span multiple hardware owners as Ian enviosioned, never the less numerous unrelated applications of arbitrary complexity belonging to unassociated users all run on the same grid. We specifically designed AppLogic so that it not only had no proprietary APIs, but in fact requires no new APIs at all. And lastly, while many customers use the grids for small applications, there are plenty of sizeable applications running on grids as well. In essence, while we’ve stretched the market for grid and intend to keep moving that boundary, we certainly haven’t broken the original spirit of the grid model.
Judging AppLogic from the information you have (ie, having never tried it or run any other grid offering) is a bit like a plumber giving you an opinion on buying a Ferrari based on looking at the dealer’s clogged sink. You just have to take it for what it’s worth.
November 30th, 2006 at 10:28 pm
Bert,
I’ll take you anology and lets apply it elsewhere.
Since many of the guys at NASA haevn’t been into outer space they aren’t qualified to talk about it?
Since a geologist has never seen the center of the earth he/she shouldn’t be able to talk about him?
Bert/Isabel/3tera,
I don’t have time to discuss with your entire team individually what a grid is. The common definition of a grid is IBM’s definition which you do not fit. You can go ahead and try and redefine grid to fit what your software can do or you can go ahead and retry to code your software to fit the proper definition of the word grid computing. It looks like you have chosen the easy marketing way.
Until your software can use the processing capabilities of multiple nodes to solve a single problem it is not a grid and you very well know it, it is a cluster.
http://en.wikipedia.org/wiki/Computer_cluster
http://www.google.com/search?q=what+is+grid+computing
December 1st, 2006 at 12:49 am
It’s funny how a host can provide “marketspeak” and all of the buzz words (which aren’t even correct for what they are doing), and then badly fail at delivering the product.
Scalability, reliability and performance of a cluster can be exclusive of each other. Meaning if you build a cluster to have scalability, you can actually make it less reliable. This is what is happening in MT setup. With 98.9% and 99.5% uptimes, their “grid” setup is actually less reliable then their previous single server Cpanel setup. It obvious they still have single points of failure.
IMHO, most hosting customers want reliability and performance first. Unlike MT marketing info most web sites are NOT slashdotted or dugg and scalability is something that usually can be planned for.
December 1st, 2006 at 2:09 am
Sorry, I took your tagline at face value; “Finding the best hosts so you don’t have to.”
December 1st, 2006 at 5:31 am
I too agree the term “grid” is widely used on these examples, the reality is that this is not grids as much as they are indeed clusters.
I am quite intereste on 3TERA’s technology, but it will be a little more testing before I can vouch for it for sure.
As much as I like the idea of splitting the services into load balanced VPS’s .. and then have a failover solution where the VPS iteself can be restored in minutes in the event of a major failure or hack of the VPS, the fact remains that by deployng VPS’s mean you are able to deploys LESS sites on the same amount of hardware, as the VPS themselves will consume large resources.
So, you get reliability at a cost of hardware .. it’s a fair and expected trade-off .. but people still expect one to charge the same $10 /mo .. thats whats unrealistic also.
December 1st, 2006 at 8:06 am
why does “titanic” spring to mind
December 1st, 2006 at 8:38 am
Hello All,
You might want to check out Carlos’s article here:
http://nullmind.com/2006/11/21/the-new-hosting-fab-grid/
December 1st, 2006 at 11:56 am
Guys, grid/cluster hosting don’t have a set meaning, they have a ruff meaning and then it depends on who you are talking to, engineers, customers, hosting companies, academics, etc…. Don’t rip on people over buzz word crap, we live in a marketing world and the truth is what we should be looking at is the exciting new technology and what it does.
And so far applogic looks very stable and well developed.
December 1st, 2006 at 3:25 pm
It is not a grid. As many have stated with a grid you expect to run a single application across many dedicated nodes rather than 10 different applications across as many servers. Since the AppLogic system is simply utilizing Xen if you have 10 apache programs as part of your “application” that is then 10 more servers you need to continuously update and monitor for security vulnerabilities, etc.
I see the AppLogic solution as a niche (as VPS is a niche) in the hosting market, not the next big step that anyone would be a fool to miss out on as I’ve read on some blogs. Most people that actually require a cluster would do better to have a few servers setup in a standard load balanced/cluster scenario.
December 2nd, 2006 at 4:20 pm
1.) That is incorrect. The appliances can be loaded off snapshots, so you can just update 1 apache appliance and then use that snapshot on all the load balancers from my reading.
2.) Grid/Cluster is a marketing term, there are lots of definitions that float around out there that could call it either. Lets just call it something we all agree on like a virtual hosting environment with redundancy.
December 4th, 2006 at 6:26 pm
Take a look here
http://www.mediatemple.net/weblog/2006/12/04/grid-improvements-new-gmr-release/
December 5th, 2006 at 4:36 am
[...] UPDATE: (mt) and Corporate Responsibility. Avlonstar’s view of the fall of the grid. Sounds more like customer loyalty than Corporate Responsibility. As a new customer I feel (mt) has handled this situation well, and reasoning that they could have ignored the situation does not make sense, especially when the grid is being advertised as the next big thing in hosting. As a new customer, satisfaction is the key, so far I have given (mt) the benefit of the doubt, but it things get worse, I am off in search of the holy grail of hosting! [...]
December 9th, 2006 at 3:03 am
[...] Technically I wasn’t kicked off, I was asked to leave the Grid Server from Media Temple. All for criticizing the grid servers lack of guaranteed uptime and performance; I later found out here that the grid isn’t a grid at all, explaining the constant outages. Problems that I noticed and brought up in the beta program months ago, all constructively and civilized. Fast forward and I’m pissed, I posted here, here, and here. The later of those posts was something that really got me on the (mt) “radar”, most likely because of it being one of the top results on Google for “media temple grid”. I had later posted some comments about the commenters of this (mt) post which were moderated, I asked why and this is what I had in my e-mail box later that day. Dan, [...]
December 11th, 2006 at 11:27 pm
Check out the new article:
http://www.digg.com/tech_news/Why_live_life_when_you_can_buy_one
December 12th, 2006 at 12:47 am
@Isabella
Why on god’s green earth would Ross PAY to support 3tera and TGL, when he feels they’re being dishonest to customers? As he has said, and I fully agree with, you don’t need to USE applogic to see that it isn’t a grid. You can try and talk around what he’s saying till you’re blue in the face, but it won’t change the fact that the way you’re distributing resources is NOT a grid.
The whole innocent victim being picked on by big bad Ross the bully thing is NOT working for you, I suggest taking another approach. For instance, if you’d like to argue why you feel the technical aspects of applogic make it a grid, please do so. If your only response is “you can’t possibly know it’s not a grid without spending money and supporting us”, you aren’t going to get far with anyone of even a basic technical background.
@Ross
Keep on fighting the good fight. I’m attempting to get some honest answers out of mr. vlad on WHT as well, but he seems to like to cherry pick responses, and sidestep questions that may show the amazing grid in bad light. The current implementation is NOT the holy grid they claim it to be.
December 20th, 2006 at 6:04 pm
Hi,
I’m a LayeredTech TGL client (one of the early ones) and I’m regretting ever moving to their grid.
I’ve experienced more downtime in the last few weeks than I had in the last few years of running on dedicated servers. 3tera’s Applogic software is not stable. Looks like MT’s service is no better:
http://chris.pirillo.com/2006/11/29/media-temple-mt-grid-server-outage/
However, MT is orders of magnitude cheaper $20 @ MT vs $200 @ TGL for 100GB of hosting space.
All-in-all looks like both sets of companies are overpromising and under-delivering with early adopters like me paying the price.
April 14th, 2007 at 4:28 am
Iam on the MT grid server. lots of downtime and latency up to 25 sec. I will move out to another hosting company.
August 9th, 2007 at 8:44 pm
gird also lame here in europe, latency time is up to 10 sec more. real slow